Exploring repetitive negative thoughts (RNT) in the Brazilian population: a correlational analysis

Full reference: Leão, C. S., Ruiz, F. J., Kovac, R., & McHugh, L. (2025). Explorando pensamentos negativos repetitivos (PNR) na população brasileira: uma análise correlacional. Perspectivas em Análise do Comportamento, Special issue: Terapia de Aceitação e Compromisso, 2025, pp. xxx–xxx. doi: 10.18761/PAC.ACT.006

Study type: Cross-sectional correlational study with mediation analysis

Background and objectives

Acceptance and Commitment Therapy (ACT) is a transdiagnostic psychological treatment model that aims to develop psychological flexibility (PF). Psychological inflexibility (PI) involves lack of contact with the present moment, rigid verbal control (cognitive fusion and self-as-content), avoidance of negative private experiences (experiential avoidance), and lack of clarity and distancing from personal values. From the ACT perspective, repetitive negative thoughts (RNT) are a type of experiential avoidance strategy composed of recurrent, aversive, and intrusive thoughts that encompass the processes of rumination and worry. RNT serves three apparent functions: avoidance of unpleasant internal experiences, emotional regulation, and problem-solving. However, when they become an inflexible pattern, RNT becomes a clinical problem. Cognitive fusion (CF) is another component of PI that influences counterproductive patterns of RNT, leading the person to behave as though their thoughts were the absolute truth in a given situation.

Previous research on RNT and ACT had been conducted with Spanish-speaking populations, but no studies had assessed the correlations between RNT and aspects of PI in the Brazilian population. Considering this, the present study had two objectives: first, to analyze the relationships between PI, CF, and psychological distress (PD) with RNT in the Brazilian sample; and second, to explore whether the relationship between PI levels and repetitive negative thinking has cognitive fusion as a mediator.

Method

Participants

Participants were 225 Brazilian adults (aged 18 and over) recruited through social media (Facebook, Instagram, and WhatsApp) during June 2020. Of the total, 76% were male (n = 171), 23.6% female (n = 53), and 0.4% identified as non-binary (n = 1). Regarding age distribution: 30.7% were aged 18-25 (n = 69), 22.7% aged 26-30 (n = 51), 16% aged 31-40 (n = 36), 14% aged 41-50 (n = 32), 11.6% aged 51-60 (n = 26), and 4.9% over 60 years (n = 11). Of the 225 participants, only 187 completed the questionnaire measuring psychological distress (DASS-21).

Design

Cross-sectional correlational study with a mediation analysis component. Two blocks of correlational analyses were conducted: the first with 225 participants (without the psychological distress variable) and the second with 187 participants (including the DASS-21).

Outcome measures

Acceptance and Action Questionnaire – II (AAQ-II; Barbosa & Murta, 2015**)**: 7-item Likert scale (1 = never true to 7 = always true) measuring psychological inflexibility. Higher scores indicate greater PI levels. Internal consistency in this study: Cronbach's α = 0.936.

Depression, Anxiety and Stress Scales – 21 (DASS-21; Antony et al., 1998; Brazilian version: Vignola & Tucci, 2014**)**: 21-item Likert scale (0 = does not apply to me to 3 = applies very much to me) measuring negative emotional states. Internal consistency in this study: total α = 0.96; subscales: depression α = 0.911, anxiety α = 0.898, stress α = 0.906.

Perseverative Thinking Questionnaire (PTQ; Ehring et al., 2011; Portuguese version: Chaves et al., 2013**)**: 15-item Likert scale (0 = never to 3 = almost always) designed to assess repetitive negative thinking processes. It had not been validated for the Brazilian population and was adapted for this study. Internal consistency: Cronbach's α = 0.964.

Cognitive Fusion Questionnaire (CFQ-7; Gillanders et al., 2014; Brazilian version: de Paula et al., 2018**)**: 7-item Likert scale (1 = never true to 7 = always true). Internal consistency in this study: Cronbach's α = 0.935.

Data analysis

The minimum sample size was calculated using G*Power 3.1 (power (1-b) > 0.8; d = 0.3), yielding n = 132. Data were analyzed with IBM SPSS version 26. The Shapiro-Wilk test indicated that none of the distributions were normal (p < 0.001), so bootstrapping with 1000 repetitions (95% BCa CI) was applied. Bivariate Pearson correlations were computed along with effect sizes (r²×100) and Fisher's Z transformation. For comparing correlation magnitudes, the online tool Testing the significance of correlations was used. Mediation analysis was conducted using the PROCESS v. 3.5 extension by Andrew F. Hayes, Model 4 (simple mediation), with 5000 bootstrap samples.

Results

In the first correlational analysis (n = 225), all correlations were positive and significant. RNT scores (PTQ) were strongly correlated with PI levels (AAQ-II; r = 0.800; p < 0.001; r² = 64%; z = 1.18). A strong correlation was also found between CF (CFQ) and RNT (r = 0.830; p < 0.001; r² = 68.4%; z = 1.19). Similarly, PI levels were strongly correlated with CF (r = 0.794; p < 0.001; r² = 63.04%; z = 1.08). The difference in strength between the correlations of RNT with PI and CF was not statistically significant (z = 1.208; p = 0.114).

In the second correlational analysis (n = 187), which included psychological distress, correlations between RNT and PD were significant but weak: total PD (r = 0.230; p < 0.001; r² = 5.29%), depression (r = 0.181; p < 0.001; r² = 3.27%), anxiety (r = 0.211; p < 0.001; r² = 4.45%), and stress (r = 0.209; p < 0.001; r² = 4.36%). Correlations of PD with CF (r = 0.230; p < 0.001; r² = 5.3%) and PI (r = 0.230; p < 0.001; r² = 5.3%) were also significant but weak. No statistically significant differences were found between the effect sizes of the DASS-21 subscale correlations with CF and PI.

The mediation analysis (n = 225) revealed that without the mediation factor, PI levels explain 63.92% of RNT levels. With CF mediation, this influence increases to 73.9%. The total effect of PI on RNT was c = 1.0619, while the direct effect was c' = 0.5137. The indirect effect through CF was statistically significant (b = 0.548; 95% BCa CI [0.4243, 0.6776]), indicating that 51.62% of RNT levels related to PI are explained by CF levels. The PI→CF path showed b = 0.8036 (95% CI [0.7278, 0.8795]; t = 20.86; p < 0.001; r² = 63.06%), and the CF→RNT path showed b = 0.682 (95% CI [0.546, 0.819]; t = 9.85; p < 0.001; r² = 73.9%).

Discussion and conclusions

The results confirm that the correlations between RNT and CF, PD, and PI in the Brazilian population are consistent with findings in the literature from other contexts. The strong correlation between RNT and PI (r = 0.8) points to a possible overlap between these variables, likely because RNT frequently functions as experiential avoidance—one of the facets of the PI model. However, correlations between RNT and psychological distress were considerably weaker than in previous studies with Spanish-speaking populations. The authors suggest this may be due to the COVID-19 pandemic context (data were collected during the first wave, in July 2020) or to a possible sample bias.

The mediation analysis confirmed that CF plays a significant mediating role in the relationship between PI and RNT: the relationship is more significant when CF is presented as a mediator, explaining 73.9% of RNT levels compared to 63.92% without the mediator. This suggests that working to reduce CF—by promoting cognitive defusion—can significantly contribute to reducing RNT levels.

Among the limitations, the authors note the possible gender and age bias in the sample (76% male; 52.4% aged between 18 and 30), the COVID-19 pandemic context, and the exclusive use of self-report measures. They recommend replication of the study outside the pandemic context, with more representative samples, and the use of implicit measures or digital monitoring.

Significance and contribution

This study constitutes the first investigation mapping the relationships between RNT, psychological inflexibility, cognitive fusion, and psychological distress in the Brazilian population. By demonstrating that CF is a significant mediator in the PI–RNT relationship—explaining more than half (51.62%) of this relationship—the findings provide evidence that interventions focused on cognitive defusion could be particularly relevant for reducing RNT in this population. Furthermore, the study extends prior research conducted with Spanish-speaking populations to the Brazilian context, a low- and middle-income country where anxiety and depression diagnoses affect 15.1% of the population.

---