Cross-national measurement invariance of the Perseverative Thinking Questionnaire for clinical psychology professionals and trainees

Title

Cross-national measurement invariance of the Perseverative Thinking Questionnaire for clinical psychology professionals and trainees

Complete Reference

Ccoyllo-Gonzalez, L., Blancas-Guillen, J., Salazar-Alvarez, G., Valencia, P. D., & Ruiz, F. J. (2025). Cross-national measurement invariance of the Perseverative Thinking Questionnaire for clinical psychology professionals and trainees. Revista Latinoamericana de Psicología, 57, 89-97. https://doi.org/10.14349/rlp.2025.v57.9

Study Type

Instrumental validation study examining cross-national measurement invariance (multigroup) of the PTQ-CPT in three Latin American countries. Total N = 475 participants (Colombia n=101; Mexico n=253; Peru n=121).

---

Context and Objectives

Clinical psychology professionals and trainees experience significant occupational stress resulting from exposure to client suffering, management of complex cases, and pressure to maintain effective therapeutic alliances. This exposure can lead to Repetitive Negative Thinking (RNT), manifesting as persistent doubts about professional competence, worries about clinical decisions, and questioning of intervention effectiveness. RNT is conceptualized as a transdiagnostic process—a common risk factor for emotional disorders such as anxiety and depression—and is associated with greater emotional exhaustion, lower job satisfaction, impaired cognitive functioning, decision-making deficits, and reduced clinical practice efficacy.

The Perseverative Thinking Questionnaire for Clinical Psychology Trainees (PTQ-CPT) was developed by Dereix-Calonge et al. (2019b) as a specific adaptation of the original PTQ (Ehring et al., 2011) to measure RNT in this population. Although prior studies demonstrated the PTQ-CPT to be a promising tool, no comprehensive psychometric reexamination has been conducted since 2019, and the equivalence of its functioning across different Latin American cultural contexts remained unclear.

Objective: To examine the psychometric properties of the PTQ-CPT and its cross-national measurement invariance in clinical psychology professionals and trainees from Colombia, Mexico, and Peru, emphasizing evaluation of whether factor structure and item functioning are equivalent across countries.

---

Method

Participants

Non-probabilistic sample of 475 participants (76.4% female; 23.6% male) distributed as follows:

• Mexico: 253 participants (53.3%)
• Peru: 121 participants (25.5%)
• Colombia: 101 participants (21.3%)

Occupational demographics:

• 24.4% undergraduate clinical psychology students
• 75.6% licensed clinical professionals
• 84% engaged in clinical interventions
• 42.2% in assessment activities
• 41.2% in health promotion and prevention
• 26.5% in triage and counseling

Weekly clinical supervision:

• 43.8% no supervision
• 41.7% 0-2 hours
• 9.3% 3-5 hours
• 5.3% more than 5 hours

Practice setting: 67.4% private sector; 17.6% public sector; 14.9% both sectors.

Sample size was determined via Monte Carlo simulation based on Dereix-Calonge et al. (2019b) parameters, indicating need for at least 100 participants per country group for measurement invariance analysis.

Instrument Evaluated: Perseverative Thinking Questionnaire for Clinical Psychology Training (PTQ-CPT)

Version and adaptation: Dereix-Calonge et al. (2019b); adapted version for present study.

Structure: 9 items measuring perseverative thinking specifically related to clinical practice. Response scale: 0–4 (from "never" to "almost always").

Modification implemented: Item 5 was modified from "I cannot concentrate on my studies" to "I cannot concentrate on other activities" for applicability to both students and licensed professionals.

Reliability in original validation: Cronbach's alpha = .93; unidimensional structure.

Prior validity evidence:

• Positive correlations with experiential avoidance, emotional symptoms, and values obstruction
• Moderate negative correlations with life satisfaction and values progress
• Longitudinal studies showed PTQ-CPT predicted emotional symptom increase better than original PTQ in clinical psychology trainees
• PTQ-CPT scores mediated effect of Acceptance and Commitment Therapy (ACT) interventions in reducing emotional symptoms

Other Measurement Instruments

Patient Health Questionnaire-2 (PHQ-2; Kroenke et al., 2003):

• Measures depressive symptoms in past two weeks
• 2 items, Likert scale 0–3
• Internal consistency in Latin American samples: ω = .86–.89 (Colombia, Mexico, Peru)
• Evidence of measurement invariance across these countries

Generalised Anxiety Disorder Scale-2 (GAD-2; Kroenke et al., 2007):

• Evaluates main anxiety symptoms in past two weeks
• 2 items, Likert response options 0–3 (0–6 total range)
• Internal consistency in Latin American samples: ω = .87–.88 (Colombia, Mexico, Peru)
• Evidence of measurement invariance across these countries

Ad Hoc Question on Confidence in Clinical Skills:

• Developed specifically for this study
• Measures perceived confidence in clinical skills as a psychologist or psychology student
• Response scale 0–10 ("not sure at all" to "totally sure")

Procedure

Data collection: Digital survey via Google Forms (March–May 2022) Coordination: Associations, institutions, instructors, and students from Peru, Mexico, and Colombia facilitated distribution through social networks Consent: Form included informed consent section explaining research objectives, voluntary nature, and anonymity Ethical approval: Ethics Committee and Research Directorate of Psychology School at Universidad Autónoma del Perú (June 26, 2023)

Data Analysis

1. Descriptive statistics: Examined at item level (mean, standard deviation, skewness, kurtosis, response option distribution).

2. Confirmatory Factor Analysis (CFA):

• Method: Robust maximum likelihood (MLR) variant
• Justification: Appropriate when at least five response options present
• Fit criteria (good fit values in parentheses): CFI > .95; TLI > .95; RMSEA < .06; SRMR < .08
• Robust versions: Robust CFI, Robust TLI, Robust RMSEA (Brosseau-Liard et al., 2012)
• If inadequate fit: examination of modification indices by country
• Reliability: omega (ω) internal consistency coefficient

3. Measurement invariance (multigroup):

• Method: Multigroup CFA with sequential equality restrictions
• Metric invariance: restrictions on factor loadings
• Scalar invariance: restrictions on intercepts
• Evaluation criteria:
  • Strict criterion: Δχ² (chi-square difference non-significant)
  • Pragmatic criterion: ΔRobust CFI ≤ .01 (Cheung & Rensvold, 2002)
• Additional procedure: Yoon and Lai (2018) subsampling with 10,000 replications to control for group size imbalance

4. Graded Response Model (IRT):

• Two-parameter model for polytomous items within item response theory framework
• Estimated parameters: discrimination (a) for each item; difficulty (b₁–b₄) for each response level
• Parameters b indicate latent construct level (θ) at which probability of selecting higher response option equals 50%
• Item information curves graphed to show θ levels where each item has greatest informative capacity

5. Validity analyses:

• Pearson correlations between PTQ-CPT scores and: depressive symptomatology (PHQ-2), anxiety symptomatology (GAD-2), confidence in clinical skills (ad hoc question)
• 95% confidence intervals

Software: R version 4.3.0; packages: lavaan (0.6-16), semTools (0.5-6), mirt (1.39)

---

Results

Descriptive Statistics at Item Level

Most items showed mean scores around 1 to 2:

Item	M	SD	Range	Highest Mean	Lowest Mean
1. Constant doubts	1.98	0.90	Highest
2. Cannot stop worrying	1.67	1.07
3. Cannot concentrate	0.95	0.97
4. Repetitive thoughts	1.28	1.07
5. Get stuck	0.81	0.91
6. Same questions	0.97	1.00
7. Think all the time	0.97	1.00
8. Do not help	1.11	1.15
9. Insecurity absorbs attention	0.79	0.95	Lowest

Most distributions displayed mild to moderate positive skewness (range g₁ = .04 to 1.15), consistent with clinical populations exhibiting moderate-to-low RNT. Kurtosis ranged from -.62 to .98.

Unidimensional Confirmatory Factor Analysis

Initial model (unmodified):

• χ²(27) = 273.56, p < .001
• CFI = .95; Robust CFI = .95
• TLI = .93; Robust TLI = .95
• RMSEA = .10; Robust RMSEA = .11
• SRMR = .04

Evaluation: Adequate fit by most approximate fit indices except RMSEA.

Modification indices examination: Indicated need to permit residual covariation between items 1 and 2 (both related to intrusive nature of thoughts).

Final model (with error correlation between items 1–2):

• χ²(26) = 234.20, p < .001
• CFI = .98; Robust CFI = .98
• TLI = .97; Robust TLI = .97
• RMSEA = .07; Robust RMSEA = .07
• SRMR = .03

Conclusion: Unidimensional structure designated as final model.

Internal reliability: ω = .92 (excellent)

Standardized factor loadings (final model):

• Item 1: .65
• Item 2: .73
• Item 3: .83
• Item 4: .83
• Item 5: .87 (highest)
• Item 6: .77
• Item 7: .70
• Item 8: .81
• Item 9: .81

Measurement Invariance Across Countries

Traditional approach:

Model	χ²	df	p	Robust CFI	Δχ²	Δdf	p(Δχ²)	ΔCFI
1. Configural	163.28	78	<.001	.97	-	-	-	-
2. Metric	179.68	94	<.001	.97	14.46	16	.565	.001
3. Scalar	214.63	110	<.001	.96	35.31	16	.004	-.007

Yoon and Lai (2018) approach with 10,000 replications:

Model	χ²	df	p	Robust CFI	Δχ²	Δdf	p(Δχ²)	ΔCFI
1. Configural	136.71	78	<.001	.97	-	-	-	-
2. Metric	153.15	94	<.001	.97	15.52	16	.487	0
3. Scalar	180.79	110	<.001	.96	27.77	16	.034	-.006

Invariance conclusions:

• Metric invariance: CONFIRMED (p > .05 both approaches; ΔCFI < .01). Factor loadings do not differ significantly across countries.
• Scalar invariance:
  • Pragmatic criterion (ΔCFI < .01): SUPPORTED (ΔCFI = -.007 traditional approach; -.006 Yoon-Lai)
  • Strict criterion (Δχ²): NOT SUPPORTED (p = .004 traditional approach; p = .034 Yoon-Lai)

Interpretation: Authors concluded scalar invariance achieved according to pragmatic criterion more commonly used, although strict criterion suggests some intercept variance across countries.

Graded Response Model (IRT)

Item discrimination parameters (a):

Item	a	Interpretation
5. Get stuck	3.97	Best discrimination
4. Repetitive thoughts	3.26
7. Think all the time	2.91
9. Insecurity	2.97
6. Same questions	2.64
2. Worry	2.41
3. Concentrate	3.15
8. Do not help	2.19
1. Doubts	2.02	Poorest discrimination

Difficulty parameters (b) indicators:

• "Easiest" item (requires lower RNT level): Item 2 (b₁ = -1.38)
• "Most difficult" item (requires higher RNT level): Item 6 (b₄ = 2.80)

Item information curves (Figure 2):

• Item 5 demonstrated greatest informative capacity, especially at higher RNT levels (θ > 1)
• Items 2 and especially item 1 provide higher-quality measurements at lower end of construct distribution (θ < -1)
• Other items show similar patterns with informative coverage in middle range

Validity Evidence: Correlations with Other Variables

Bidirectional Pearson correlations with PTQ-CPT scores:

Variable	r	95% CI	Magnitude
Depressive symptoms (PHQ-2)	.45	[.37, .53]	Moderate positive
Anxiety symptoms (GAD-2)	.56	[.49, .64]	Moderate-to-high positive
Confidence in clinical skills	-.58	[-.64, -.52]	Moderate-to-high negative

Interpretation: Higher RNT scores associate with greater depressive and anxiety symptomatology, and with lower confidence in clinical abilities. These association patterns are theoretically expected and support PTQ-CPT construct validity.

---

Discussion and Conclusions

Main Findings

The study demonstrated that the PTQ-CPT possesses a unidimensional structure with a pair of correlated errors providing best fit to data. This structure remains consistent across three Latin American countries (Colombia, Mexico, Peru). Results are congruent with original Dereix-Calonge et al. (2019b) validation finding unidimensional structure with good fit and Cronbach's alpha > .90.

Primary difference in present study was need to permit correlated errors between items 1 and 2. While initially seeming undesirable, the fact that this modification emerged independently across all three countries indicates robustness and replicability. Conceptual justification exists: both items reference intrusive nature of thoughts.

Cross-National Measurement Invariance

Confirmation of metric invariance indicates PTQ-CPT factor loadings are equivalent across countries. This supports that the fundamental construct (RNT related to clinical practice) is measured equivalently in Colombia, Mexico, and Peru.

Scalar invariance, while not achieving strict criterion (Δχ²), was supported by pragmatic criterion widely used in literature (ΔCFI < .01). This suggests mean comparisons across countries may be interpreted with some caution, though intercept variance across countries warrants consideration.

Item Information Capacity

Item 5 ("I get stuck in my worries about the clinical practice and I cannot concentrate on other activities") demonstrated best discrimination (a = 3.97) and greatest informative capacity at elevated RNT levels. This item is especially useful for differentiating among individuals with high RNT. Items 1 and 2 provide better information at lower extreme, suggesting complete construct coverage across range.

Construct Validity

Expected correlations with depression (r = .45), anxiety (r = .56), and clinical confidence (r = -.58) were found, confirming construct validity. RNT measured by PTQ-CPT relates positively to negative emotional symptoms and negatively to professional confidence—patterns expected from repetitive thinking literature in clinical populations.

Limitations

1. Convenience sample: Non-probabilistic sampling may limit generalization to other populations or settings.
2. Self-selection bias: Voluntary participation could introduce biases, particularly related to technology access and internet availability.
3. Self-report measures: Susceptibility to common method bias and subjective interpretation (PTQ-CPT, PHQ-2, GAD-2).
4. Cross-sectional design: Precludes causal inference; permits no establishment of temporal or predictive relationships between RNT and other variables.
5. Absence of clinical diagnosis: Depressive and anxiety symptoms measured via self-report, not structured clinical diagnostic assessment.
6. Ad hoc measure: Confidence in clinical skills question was not previously validated.

Significance and Contribution

This study makes an important contribution to the field of cross-cultural measurement of repetitive negative thinking in Latin American clinical psychology professionals. The PTQ-CPT demonstrated equivalent functioning across three countries (Colombia, Mexico, Peru), validating its use in cross-cultural research and clinical practice contexts in the region. The demonstration of metric and scalar invariance across different national contexts is methodologically important, allowing valid comparisons of repetitive negative thinking scores between professionals in different settings. The study contributes to the field of Latin American clinical psychology by providing a valid and reliable tool for assessing this transdiagnostic process in clinical psychology professionals and trainees, facilitating both research and clinical application in Spanish-speaking populations.

---

Final Conclusions

The PTQ-CPT demonstrated adequate psychometric properties and cross-national measurement invariance in clinical psychology professionals and trainees from Colombia, Mexico, and Peru. The instrument showed:

• Excellent internal consistency (ω = .92)
• Replicable unidimensional structure
• Significant correlations with theoretically relevant variables
• Equivalent functioning across countries by pragmatic invariance criterion

It is a reliable and valid tool for assessing Repetitive Negative Thinking in Latin American clinical practice contexts, supporting use in cross-cultural research and clinical practice. Findings establish foundation for future longitudinal research on RNT implications for emotional well-being and clinical performance of training and practicing professionals.

---

---