Laiana Soeiro Ferreira1*
Simone Souza da Costa e Silva2Fernando Augusto Ramos Pontes2
José Augusto Evangelho Hernandez3Emmanuelle Pantoja Silva2Luana Conceição Queiroz2Christoph de Oliveira Käppler4Jonas Carvalho e Silva5- 1Developmental Ecology Laboratory, Graduate Program in Theory and Research of Behavior and Faculty of Physiotherapy and Occupational Therapy, Federal University of Pará (UFPA), Belém, Pará, Brazil
- 2Graduate Program in Theory and Research of Behavior, Center for Theory and Research of Behavior, Federal University of Pará (UFPA), Belém, Pará, Brazil
- 3Department of Psychology, Cognition and Development, State University of Rio de Janeiro (UERJ), Rio de Janeiro, Brazil
- 4Faculty of Rehabilitation Sciences, Technical University of Dortmund (TU Dortmund), Dortmund, Germany
- 5Graduate Program in Theory and Research of Behavior, Federal University of Pará (UFPA), Belém, Pará, Brazil
Stress is a natural reaction to change or challenge and can lead to various physical, emotional, and behavioral responses. It becomes problematic when stressors persist without relief or relaxation. Academic stress is interpreted as a physiological, emotional, cognitive, and behavioral trigger originating from stimuli and events within the university environment. This study sought evidence of validity and reliability for the 10-item Perceived Stress Scale (PSS-10) based on the scores of 283 university students of both genders from the Northern region of Brazil. The instrument aims to measure the degree to which individuals perceive situations as unpredictable, uncontrollable, and overwhelming. The scale consists of 10 items—six negatively worded and four positively worded. Since its development, the internal structure of the instrument has remained a subject of debate in the literature. In this study, the most frequently tested models in international research were evaluated using Structural Equation Modeling with Confirmatory Factor Analyses. The results showed that the two-factor oblique model (Perceived Stress and Perceived Coping Ability) demonstrated the best fit to the empirical data among the models tested. Nevertheless, the internal consistency value for the Perceived Stress factor fell below the adequate cutoff point. Additionally, the scale scores showed significant positive correlations with the Beck Anxiety Inventory. Overall, despite the limitations of a non-probabilistic, convenience sample, acceptable evidence of internal structure validity, correlation with an external related measure, and reliability was obtained for the scale based on scores from university students in Northern Brazil.
Introduction
The World Health Organization (World Health Organization, 2023) has identified stress as an epidemic of the twenty-first century. According to Selye (1956), stress is defined as the organism's effort to adapt when facing situations perceived as threatening to life or internal balance. This led to the concept of the General Adaptation Syndrome, which, as described by Lipp and Romano (1987), refers to the set of physical and psychological reactions experienced by the body when exposed to stimuli that excite, irritate, frighten, or even elicit intense happiness. Stress may be understood as an arousal response that can emerge across diverse contexts—family, education, work, among others—sometimes protecting life, but also potentially leading to physical and mental dysfunctions or illnesses.
Academic stress is interpreted as a physiological, emotional, cognitive, and behavioral trigger that arises from events and demands within the university environment. Uncertainties and anxieties related to assignments, exams, social pressures, and a heavy workload are commonly viewed as potential stressors. These factors contribute to negative manifestations of stress, which can lead to substantial cognitive impairment and, consequently, reduced performance in more complex tasks (Igra, 2024; Murakami et al., 2024).
Given the pervasive incidence of stress across contexts—and particularly in academic settings—the use of psychometric instruments that allow for precise and reliable assessment of perceived stress levels becomes essential. Such instruments are crucial for diagnostic and screening purposes in clinical practice and academic research, as well as for informing interventions and policies that promote mental health. Employing measures with consistent evidence of validity and reliability is vital to ensure that the results obtained are effective for health promotion in diverse cultural and social contexts (Seabra-Santos et al., 2021).
Although the present study did not stratify the sample based on characteristics related to disability, it is important to acknowledge that academic stress disproportionately affects historically marginalized groups. Research by Barbosa and Costa (2018) and Cardoso and Freitas (2020) shows that university students with disabilities face structural, social, and institutional barriers, such as limited accessibility, challenges in curricular adaptation, and insufficient psychosocial support. These conditions heighten their exposure to academic stressors and may jeopardize their retention in higher education.
This need becomes even more evident in understudied contexts, such as that of university students in the Northern Region of Brazil. This group faces unique vulnerabilities, including inequalities in access to public policies, economic barriers, and structural limitations in educational and health services (Souza et al., 2025). Furthermore, cultural and regional factors directly influence the academic and emotional experiences of these students, highlighting the necessity for context-sensitive analyses (Mattos et al., 2020). It is therefore imperative to expand the body of knowledge on the use and validation of psychological instruments tailored to this population's specific characteristics.
The Perceived Stress Scale (PSS) was developed by Cohen et al. (1983) to measure the extent to which individuals perceive their lives as unpredictable, uncontrollable, and overloaded. Its construction stemmed from the observation that objective measures of life stressors—such as counts of negative events—were insufficient to capture the subjective experience of stress. The original version, known as the PSS-14, consists of 14 items. To facilitate its use in contexts requiring shorter measures, two abbreviated versions were developed: the PSS-10, with 10 items, and the PSS-4, with four items (Cohen and Williamson, 1988). The PSS-10 is widely used because it balances brevity with strong psychometric properties, whereas the PSS-4, though practical, shows lower internal consistency.
The internal structure of the PSS has been widely debated since its creation. Although the scale was initially designed to measure perceived stress as a single global construct, with both negative and positively keyed items (the latter reverse-scored) contributing to a single total score, early conceptualizations treated the PSS as unidimensional (Cohen et al., 1983; Cohen and Williamson, 1988).
Machado et al. (2014) conducted an Exploratory Factor Analysis (EFA) using scores from elementary and high school teachers, revealing a one-factor solution with internal consistency of 0.80. Similar findings were reported by Trigo et al. (2010).
In contrast, Perera et al. (2017) and Jatic et al. (2023) performed reliability assessments and both exploratory and confirmatory factor analyses. They tested three widely discussed PSS-10 models—unidimensional, two correlated factors, and bifactor—among family health professionals in Bosnia and Herzegovina. Findings from Reis et al. (2017) supported the bifactor model for the PSS-10.
In Brazil, Reis et al. (2010) evaluated the PSS-10 with a sample of Brazilian university professors. The Portuguese version was produced through a two-step translation process. A Principal Component Analysis with Varimax rotation extracted two factors. Cronbach's alpha coefficients were 0.83 for Factor 1 (negative items) and 0.77 for Factor 2 (positive items). A subsequent Confirmatory Factor Analysis tested oblique two-factor and hierarchical models, with the hierarchical model—featuring a global second-order factor and two first-order factors—yielding the best fit.
However, factor analyses conducted in multiple countries and contexts have increasingly shown that positive and negative items tend to cluster separately, indicating the presence of two distinct factors (Chen et al., 2021; Hore-Lacy et al., 2024; Faro, 2015; Liu et al., 2020; Makhubela, 2022; Marakshina et al., 2024; Mendis et al., 2023; Mondo et al., 2021; Reis et al., 2010; Ribeiro et al., 2024; Ruisoto et al., 2020; She et al., 2021; Xiao et al., 2023).
Mondo et al. (2021), for example, analyzed the Italian versions of the PSS-14, PSS-10, and PSS-4 in a sample of workers. Two-factor models demonstrated better fit than unidimensional ones, with high reliability (Cronbach's alpha of 0.85 for the PSS-14 and 0.84 for the PSS-10).
She et al. (2021) tested the psychometric properties of all three PSS versions in a nationally representative Chinese sample during the COVID-19 pandemic. Data were collected remotely from 1,133 adults. Analyses revealed good fit for the two-factor solution, and the PSS-10 showed strong reliability.
In Portugal, Ribeiro et al. (2024) assessed the 10-item PSS in 1,369 social workers, most of them women. Principal Component Analysis with Varimax rotation revealed two factors. Confirmatory Factor Analysis indicated good fit for the two-factor model, and Cronbach's alpha results supported adequate internal consistency.
The most robust and recent evidence supporting the scale's bidimensional structure, however, comes from Kogar and Kogar (2024). They performed a systematic review and a Meta-Analytic Confirmatory Factor Analysis (MACFA) for the PSS-10 using data from 76 samples totaling over 46,000 participants. Five factor models—including unidimensional, bidimensional, and bifactor models—were compared. The two correlated-factor model demonstrated the best fit indices. Their findings also revealed measurement invariance across age groups and clinical status, as well as strong internal consistency (McDonald's omega). This study reinforces the psychometric soundness of the PSS-10 and recommends the adoption of the two-factor model in studies assessing perceived stress in diverse cultural and clinical populations.
One explanation for the variability in results—and for the different labels used to name the PSS-10 factors—may lie in method effects. According to DiStefano et al. (2022), method effects occur when variance associated with the measurement method becomes embedded in test responses. These effects may stem from various sources, such as acquiescence, social desirability, or the inclusion of positively and negatively worded items, as is the case with the PSS-10. Statistical techniques such as bifactor modeling could help detect these method effects by distinguishing them from the main construct. However, prevailing evidence for the PSS-10 indicates the presence of two distinct, correlated constructs rather than the single construct originally proposed (Cohen et al., 1983; Cohen and Williamson, 1988). The labels used for these two factors vary across studies: “Perceived Helplessness and Perceived Self-Efficacy” (Chen et al., 2021; Kogar and Kogar, 2024; Liu et al., 2020; Marakshina et al., 2024; Ruisoto et al., 2020; Xiao et al., 2023), “Negative and Positive Statements” (Reis et al., 2010), or “Perceived Stress and Perceived Coping Ability” (Jatic et al., 2023; Ribeiro et al., 2024).
Widely accepted within the scientific community, the PSS is recognized for its broad intercultural applicability. Its use has proven effective for assessing chronic stressors, future expectations, and psychosocial conditions often overlooked by other scales (Kogar and Kogar, 2024).
National and international findings underscore the relevance of the PSS-10 as a reliable instrument for evaluating perceived stress, motivating continued investigations across diverse populations. In Brazil, the first adaptation of the PSS-14 was conducted by Luft et al. (2007) with older adults in Aracaju, in the Brazilian Northeast. Reis et al. (2010) examined the PSS-10 in university professors in the South, and Faro (2013) evaluated graduate students from all regions of the country. However, no study to date has performed a Confirmatory Factor Analysis of the PSS-10 with undergraduate university students from the Northern Region of Brazil. By seeking psychometric evidence for the PSS-10 in an underrepresented region, this study contributes to strengthening the scientific foundation needed to promote more inclusive, accessible, and diversity-aware university environments.
This study sought evidence of validity and reliability for the measure of stress. To this end, the internal structure of the scale was examined through Confirmatory Factor Analysis; its relationship with an external measure—the Beck Anxiety Inventory—was assessed; and its internal consistency was evaluated using Cronbach's alpha and McDonald's omega.
Method
Participants
The study sample consisted of 283 university students (men and women) enrolled in undergraduate programs at a Higher Education Institution located in Northern Brazil. All participants were 18 years of age or older. The sample was non-probabilistic and convenience-based. Table 1 presents the main sociodemographic characteristics of the sample.
Table 1. Sociodemographic profile of participants.
The courses with the highest number of respondents were Occupational Therapy, Physical Therapy, and Medicine, all within the field of health. It was also observed that only 16.25% of the students received some type of financial assistance, with institutional academic scholarships and permanence aid being the most frequently mentioned sources. Regarding mental health, about half of the participants (48.76%) reported having sought professional support for emotional issues, such as from occupational therapists, psychologists, and/or psychiatrists.
Instruments
Sociodemographic inventory
The Sociodemographic Inventory was used to collect relevant information for the study. The following variables were investigated: gender identity, sexual orientation, race/color, religion, university campus, receipt of financial aid, and history of mental health care, including the type of emotional issue reported and the professional(s) consulted.
Perceived Stress Scale-10 (PSS-10)
The 10-item version of the PSS (Cohen and Williamson, 1988), translated into Brazilian Portuguese by Reis et al. (2010), was used in this study. In Reis et al. (2010), the instrument was translated using the back-translation method. A panel of experts then verified the theoretical coverage of the construct, which was composed of 10 items distributed across two factors: six negatively worded and four positively worded items. Brazilian university professors responded to the instrument using a five-point Likert scale (“0 = never” to “3 = always”). The internal consistency values, as measured by Cronbach's alpha, were 0.83 for factor 1 and 0.77 for factor 2. Exploratory Factor Analysis showed two factors with eigenvalues greater than 1.0 (explaining 56.8% of total variance). The Cronbach's alpha coefficients were 0.83 (Positive Statements) and 0.77 (Negative Statements). Correlations between the PSS-10 and a measure of perceived health ranged from −0.22 to −0.35. Reis et al. (2010) concluded that the PSS-10 presented adequate validity and reliability, supporting its use among Brazilian university professors. In the present study, the factors were labeled “Perceived Stress” and “Perceived Coping Ability,” as these terms functionally and succinctly reflect the dimensions of the PSS-10 and align with central principles guiding the understanding and assessment of stress in the mental health field.
Beck Anxiety Inventory (BAI)
The BAI, a unidimensional measure developed by Beck et al. (1988), consists of 21 items that assess cognitive and somatic symptoms of anxiety, rated on a Likert scale from 0 to 3, with a total score ranging from 0 to 63. This study used the Brazilian version validated by Cunha (2001), widely used in both clinical and research settings, with an internal consistency of α = 0.92. A new national version of the BAI was published in 2023 by Hogrefe Publishing, available exclusively online. Since data collection for the current study began in 2022, the Cunha (2001) version was maintained. The scores collected using the PSS-10 were correlated with the BAI; although the instruments represent different constructs, they are theoretically and empirically related.
Data collection procedures
In accordance with Resolutions 466/2012 and 510/2016 of Brazil's National Health Council, this study was submitted to and approved by the institution's ethics committee (CAAE: 65460822.0.0000.5172, Opinion No. 6.041.938). An Informed Consent Form was read and signed by all participants. The research was conducted remotely via an online questionnaire hosted on Google Forms®, from 2022 to 2024.
Dissemination was supported by course coordinators and academic centers of a public university in Northern Brazil, as well as through social media and institutional email lists. Students accessed the research link, read the informed consent form, and upon agreeing to participate, responded anonymously. The average completion time was 8 min.
Data analysis procedures
Procedures followed the guidelines established by the American Educational Research Association, et al. (2014) and the International Test Commission (2017) for cross-cultural adaptation of psychometric instruments: evidence of content validity, internal structure validity, and validity through external measures. Since the Brazilian Portuguese version of the PSS-10 by Reis et al. (2010) was used, content validity evidence was already established.
Descriptive statistical analyses were conducted to examine score distributions. A Confirmatory Factor Analysis (CFA) was performed using JASP software (JASP Team, 2025), version 0.19.3, with the Robust Diagonally Weighted Least Squares (RDWLS) estimation method, suitable for ordinal data and based on a polychoric correlation matrix. The goal was to test the most commonly proposed theoretical models for the PSS-10 structure, as identified in the meta-analysis by Kogar and Kogar (2024). Model fit quality was assessed using widely accepted statistical indices: Chi-square (χ2): measures discrepancy between observed and estimated data; CFI (Comparative Fit Index): values >0.90 indicate good fit; >0.95 excellent fit (Marôco, 2021); TLI (Tucker-Lewis Index): similar to CFI, adjusts for model complexity; SRMR (Standardized Root Mean Square Residual): values < 0.08 indicate good fit; RMSEA (Root Mean Square Error of Approximation): values < 0.05 indicate excellent fit (Hair et al., 2019).
Internal consistency was evaluated using Cronbach's Alpha and McDonald's Omega coefficients. Values above 0.70 were considered to indicate coherent construct measurement (Hair et al., 2019). Additionally, validity evidence for the PSS-10 was examined through its relationship with the BAI. Although representing different constructs, the PSS-10 and BAI are both theoretically and empirically linked. Spearman's rho correlation coefficient was used to examine the strength and direction of this association, suitable for ordinal data.
Results
Descriptive analysis revealed that the score distributions did not conform to standard statistical normality. However, skewness (between −0.796 and 0.185) and kurtosis (between −0.684 and 1.292) values were considered acceptable for multivariate analyses, according to Tabachnick and Fidell (2019). Multivariate normality was assessed using Mardia's index (Mardia, 1970), which confirmed significant multivariate skewness (value = 138.61; CR index = 10.54). No outliers were identified.
Using the RDWLS estimation method, suitable for Likert-type ordinal responses (Li, 2016), four theoretical models of the PSS-10 internal structure were tested: Unidimensional, two correlated (oblique) factors, hierarchical (two first-order factors + one second-order general factor) and Bifactor.
The unidimensional model did not show satisfactory fit. In contrast, the hierarchical and two oblique factors models showed very good fit (Table 2), according to Marôco (2021).
Table 2. Overall fit indices of the tested PSS-10 models: previous and current.
The bifactor model also demonstrated acceptable fit, but its derived indices did not support a strong general factor along with two independent specific factors. The following complementary indices were considered: Explained Common Variance (ECV)—proportion of common variance explained by the general factor; Percent of Uncontaminated Correlations (PUC), proportion of item correlations reflecting only the general factor. According to Rodriguez et al. (2015), values above 0.70 for both indices suggest essentially unidimensional structure. In this study, ECV = 0.47 and PUC = 0.53, which do not meet this criterion. These results reinforce the interpretation that the two specific factors reflect distinct and non-redundant dimensions of the perceived stress construct.
To compare model fit and determine whether the two correlated factors model provided statistically superior fit, hypotheses were tested based on chi-square values (χ2) and corresponding degrees of freedom: According to the null hypothesis (H0), there would be no significant difference between the compared models; that is, the chi-square values would be equal: H1: χ2_original = χ2_respecified. The alternative hypothesis (H1) assumed that the model fits would differ significantly: H1: χ2_original ≠ χ2_respecified.
Model comparison was based on the chi-square difference (Δχ2 = χ21 – χ22) and the difference in degrees of freedom. The chi-square difference was calculated to identify the best-fitting model. Comparison between the unidimensional model and the two correlated factors model: Δχ2 = 107.756 – 67.668 = 40.088, with a difference of 1 degree of freedom (35 – 34). The critical chi-square value for α = 0.05 with 1 degree of freedom is 3.841, thus: χ20.95(1) = 3.841 < Δχ2 = 40.088. This result indicated a statistically significant difference, allowing the null hypothesis (H0) to be rejected. Therefore, it was concluded that the model with two correlated factors presented a significantly better fit than the unidimensional model (Δχ2 = 40.088, p < 0.05). This superiority was also corroborated by the lower ECVI (Expected Cross-Validation Index) value, indicating better predictive capacity of the bifactor model.
Comparison between the unidimensional model and the hierarchical model: Δχ2 = 107.756 – 67.668 = 40.088, with 2 degrees of freedom (35 – 33). The critical chi-square value for α = 0.05 with 2 degrees of freedom is 5.991, therefore: χ20.95(2) = 5.991 < Δχ2 = 40.088. Again, the null hypothesis (H0) was rejected, demonstrating that the hierarchical model also fit the data significantly better than the unidimensional model.
Although the hierarchical model showed a good fit, the two correlated factors model achieved superior fit to the data, being classified as a “very good fit” according to the criteria of Marôco (2021). Furthermore, the comparison of ECVI values between the two models also favored the two correlated factors model, confirming its greater adequacy (see Table 2).
Regarding the PSS-10 items, the standardized factor loadings of the items composing the two factors—Perceived Stress (PS) and Perceived Coping Ability (PCA)—were mostly equal to or above 0.50, a value indicating good individual item reliability, according to the criteria of Hair et al. (2019).
The only exception was item 07, whose value fell slightly below the cutoff point (see Table 3). Nevertheless, this value was close to the recommended threshold and did not compromise the structure of the factor to which it belongs.
Table 3. Items and factor loadings of the tested PSS-10 models.
Table 3 presents the standardized factor loadings of the PSS-10 items across the different models tested: unidimensional, two oblique factors, and hierarchical. Factor loadings represent the extent to which each item contributes to the factor it is associated with. In psychometric analyses, values equal to or greater than 0.50 are considered desirable, as they indicate that the item has good representativeness within the factor (Hair et al., 2019).
In the two oblique factors model, the negatively worded items (Perceived Stress) showed satisfactory loadings, ranging from 0.52 to 0.69. These results reinforce the internal consistency of this dimension. The positively worded items (Perceived Coping Ability), although showing slightly lower loadings, still remained within the acceptable range (between 0.47 and 0.60).
In the hierarchical model, similar patterns were observed, and the first-order factors (PS and PCA) continued to exhibit good internal consistency. The second-order factor (PSS) also demonstrated adequate saturation values: 0.73 for PS and 0.78 for PCA, indicating that both specific factors are well-related to the general factor proposed in the hierarchical model.
Discriminant validity refers to the ability of an instrument to distinguish between distinct concepts, that is, to the relative independence between the factors representing latent constructs (Fornell and Larcker, 1981; Marôco, 2014). To assess discriminant validity in the present sample, a comparison was conducted between the Average Variance Extracted (AVE) of each factor and the squared correlation between the PSS-10 factors.
The results demonstrated that the two factors in the model—Perceived Stress and Perceived Coping Ability—showed explained variances greater than the squared correlation value, which indicates that they are sufficiently distinct from each other and therefore possess discriminant validity (Table 4).
Table 4. Average variance extracted, correlation index, HTMT, and internal consistency of PSS factors with confidence intervals.
In addition, the Heterotrait-Monotrait Ratio of Correlations (HTMT) method, as proposed by Henseler et al. (2015), was also used as a complementary measure of discriminant validity. HTMT assesses the similarity between latent variables, with values ranging from 0 to 1, where lower values suggest greater distinction between factors. The results further supported the presence of adequate distinction between the two factors of the scale.
Finally, internal consistency coefficients for the PSS-10 factors were also calculated (Table 4).
To obtain validity evidence based on external criteria, in accordance with the guidelines of the American Educational Research Association, et al. (2014) and the International Test Commission (2017), the relationship between scores from the PSS-10 and the Beck Anxiety Inventory (BAI) was analyzed. The results indicated moderate and statistically significant correlations between the PSS-10 factors and the BAI.
Specifically, the Perceived Stress (PS) factor showed a Spearman's rho correlation of 0.57 (p < 0.001) with anxiety scores. This result suggests that the greater the level of perceived stress related to overload and lack of control, the greater the level of anxiety symptoms tends to be. The Perceived Coping Ability (PCA) factor showed a negative correlation of −0.34 (p < 0.001), indicating that perceptions of coping and control are inversely associated with anxiety levels, although with lower intensity. These findings support the convergent validity of the PSS-10 by showing that its factors are theoretically and empirically related to a closely linked psychological construct such as anxiety. Additionally, the relationship between the PS and PCA factors of the PSS-10 yielded a Spearman's rho of−0.46 (p < 0.001).
Discussion
The aim of the present study was to examine the psychometric properties of the Brazilian version of the Perceived Stress Scale (PSS-10), providing evidence of validity and reliability based on the scores of university students from Northern Brazil. Content validity evidence for this version of the PSS-10 had previously been reported by Reis et al. (2010). Preliminary analyses of the current data indicated that the variables were not normally distributed. Therefore, the items were treated as ordinal, and Confirmatory Factor Analyses (CFAs) were conducted using the Robust Diagonally Weighted Least Squares estimator with a polychoric correlation matrix, which is appropriate for this type of data.
Among the tested models, the two correlated factors model demonstrated the best fit to the empirical data. This finding is consistent with previous studies, such as Reis et al. (2010), conducted with university professors, where the two-factor model also showed an advantage over the hierarchical model. In the present study, this difference was smaller but still favored the two correlated factors model.
The meta-analytic confirmatory factor analysis conducted by Kogar and Kogar (2024) reinforced this pattern, identifying 38 studies that tested the PSS-10 structure across different countries and samples. In 32 of these studies, the two correlated factors model was the most frequently supported. However, the analytical methods used were not always the same: for example, 27 studies employed Maximum Likelihood estimation.
In the current study, nearly all items presented factor loadings equal to or above 0.50, indicating that a substantial portion of item variance was explained by the latent factors, thus supporting item-level reliability (Hair et al., 2019; Marôco, 2021). Only item 07 fell below this cutoff, but remained close to the acceptable threshold. Discriminant validity between factors was confirmed using the criteria of Fornell and Larcker (1981), as the average variance extracted (AVE) values were greater than the squared correlation between the factors. Additionally, the Heterotrait-Monotrait Ratio (HTMT) of 0.54 further supported the distinction between the two factors. According to Henseler et al. (2015), HTMT values below 0.85 indicate reliable discriminant validity among latent dimensions.
Regarding the bifactor model, this structure allows for the coexistence of a general factor alongside specific factors, each associated with a subset of items. However, complementary tests based on the Explained Common Variance (ECV) and the Percent of Uncontaminated Correlations (PUC) did not support the assumption that the PSS-10 should be considered essentially unidimensional (Markon, 2019; Reise et al., 2018; Rodriguez et al., 2015).
With respect to reliability, it is important to acknowledge the methodological limitations of traditional internal consistency estimators, such as Cronbach's alpha and McDonald's omega. Alpha can be affected by violations of tau-equivalence (Hauck-Filho and Valentini, 2020), while omega is sensitive to sample size and number of items (Edwards et al., 2021). Both coefficients showed satisfactory values in this study, meeting the ≥ 0.70 criterion suggested by Hair et al. (2019) for the EP factor; however, internal consistency values for the CEP factor were lower than expected (Table 4).
Evidence of validity based on the relationship between the PSS-10 and an external measure—the Beck Anxiety Inventory (BAI)—was also obtained. The correlation with the EP factor was moderate and significant, whereas the correlation with the CEP factor was negative and weaker. According to the International Test Commission (2017), this type of validity indicates that the scale is associated with a theoretically related construct—in this case, anxiety. The coefficients observed were within the expected range of 0.20 to 0.50 for convergent correlations (Nunes and Primi, 2010) and aligned with findings from international research (Acoba, 2024; Ebrahim et al., 2024; Liu et al., 2020; Ruisoto et al., 2020; Trigo et al., 2010; Zhang et al., 2023).
Although this study did not specifically assess students with disabilities, the results are directly relevant to contemporary discussions on inclusion in higher education. The elevated levels of perceived stress identified suggest the presence of academic conditions that may intensify barriers to participation, engagement, and wellbeing. For students with disabilities, who already encounter structural, attitudinal, and institutional obstacles, these stressors may represent an additional burden, potentially widening inequalities and compromising academic persistence (Barbosa and Costa, 2018; Cardoso and Freitas, 2020). By demonstrating that the PSS-10 functions adequately in the Northern Brazilian context, this study supports the use of reliable psychosocial indicators in institutional monitoring systems, informing the planning of support services, accessibility measures, and higher education policies aligned with inclusive principles.
Conclusion
In summary, through analyses of internal structure and its relationship with an external variable, the results provide validity evidence for the two correlated factors model of the PSS-10 in a sample of university students from Northern Brazil. The internal consistency of the Perceived Stress factor fell slightly below the 0.70 threshold, possibly suggesting a need for better adaptation of items for the target population. It is recommended that future research review the content of the items comprising this factor in the Brazilian version of the PSS-10.
By including a sample from the Northern region, the study broadens the understanding of the PSS-10's functioning across diverse cultural and geographic contexts in Brazil, contributing to the decentralization of psychometric research, which remains concentrated in the South and Southeast regions. In this regard, the findings reinforce the scale's applicability as a screening and mental health research tool, especially in academic settings where stress levels are alarmingly high.
Thus, this research contributes to the development of interventions aimed at promoting mental health in academic environments.
Although the present results offer novel contributions to the study of this measure in Brazil, its limitations must be acknowledged—particularly the non-probabilistic sample of university students, limited to the state of Pará and to certain socioeconomic strata of the population. It is suggested that future research test the PSS-10 with more diverse samples from other Brazilian regions and with varied sociodemographic profiles.
Data availability statement
The data analyzed in this study is subject to the following licenses/restrictions: the dataset used in this study is not publicly available due to ethical restrictions and the protection of participant privacy, as approved by the Research Ethics Committee of the Federal University of Pará (CAAE No. 65460822.0.0000.5172). Data may be made available upon reasonable request to the authors and subject to additional ethical approval. Requests to access these datasets should be directed to bGFpYW5hc29laXJvQHVmcGEuYnI=.
Ethics statement
The studies involving humans were approved by NÚCLEO DE MEDICINA TROPICAL DA UFPA. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.
Author contributions
LF: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. SS: Funding acquisition, Methodology, Supervision, Writing – original draft, Writing – review & editing. JH: Methodology, Writing – original draft, Writing – review & editing. ES: Writing – review & editing, Methodology, Writing – original draft. FP: Writing – review & editing. LQ: Writing – review & editing. CK: Writing – review & editing. JS: Writing – review & editing.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This research was funded by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Brazil, through the Programa de Desenvolvimento Acadêmico Abdias Nascimento – Grant Code 001. The funding agency supported the research activities but had no role in the study design, data collection, analysis, or decision to publish.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher's note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
Acoba, E. F. (2024). Social support and mental health: the mediating role of perceived stress. Front. Psychol. 15:330720. doi: 10.3389/fpsyg.2024.1330720
American Educational Research Association American Psychological Association, and National Council on Measurement in Education. (2014). The Standards for Educational and Psychological Testing. Washington, DC: American Educational Research Association.
Barbosa, B. C., and Costa, M. P. R. (2018). Inclusion in higher education: barriers and facilitators for students with disabilities. Rev. Brasil. Educação 23:e230088.
Beck, A. T., Epstein, N., Brown, G., and Steer, R. A. (1988). An inventory for measuring clinical anxiety: psychometric properties. J. Consult. Clin. Psychol. 56, 893–897. doi: 10.1037/0022-006X.56.6.893
Cardoso, A. P., and Freitas, C. R. (2020). Accessibility and retention policies for university students with disabilities in Brazil. Educação Sociedade 41:e238145.
Chen, J. Y., Chin, W.-Y., Tiwari, A., Wong, J., Wong, I. C. K., Worsley, A., et al. (2021). Validation of the Perceived Stress Scale (PSS-10) in medical and health sciences students in Hong Kong. TAPS Int. J. Health Profess. Educ. Centered Asia 6, 31–37. doi: 10.29060/TAPS.2021-6-2/OA2328
Cohen, S., Kamarck, T., and Mermelstein, R. (1983). A global measure of perceived stress. J. Health Soc. Behav. 24, 385–396. doi: 10.2307/2136404
Cohen, S., and Williamson, G. M. (1988). “Perceived stress in a probability sample of the United States,” in The Social Psychology of Health: Claremont Symposium on Applied Social Psychology, eds. S. Spacapan and S. Oskamp (Newbury Park, CA: Sage), 31–67.
Cunha, J. A. (2001). Manual da versão em português das Escalas de Beck. São Paulo: Casa do Psicólogo.
DiStefano, C., Schweizer, K., and Troche, S. (2022). Editorial: Controlling psychometric measures for method effects by means of factor analysis. Front. Psychol. 13:984050. doi: 10.3389/fpsyg.2022.984050
Ebrahim, O. S., Sayed, H. A., Rabei, S., and Hegazy, N. (2024). Perceived stress and anxiety among medical students at helwan university: a cross-sectional study. J. Public Health Res. 13:22799036241227891. doi: 10.1177/22799036241227891
Edwards, M. C., Flora, D. B., and Thissen, D. (2021). A simulation study on the performance of different reliability estimators. Psychol. Methods 26, 111–123.
Faro, A. (2013). Análise fatorial confirmatória das três versões da Perceived Stress Scale (PSS): um estudo populacional. Psicologia 28, 21–30. doi: 10.1590/1678-7153.201528103
Faro, A. (2015). Análise fatorial confirmatória das três versões da Perceived Stress Scale (PSS): um estudo populacional. Psicol. Reflexão Crítica 28, 21–30. doi: 10.1590/1678-7153.201528103
Fornell, C., and Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. J. Market. Res. 18, 39–50. doi: 10.1177/002224378101800104
Hair, J. F. Jr., Black, W. C., Babin, B. J., and Anderson, R. E. (2019). Multivariate Data Analysis, 8th Edn. Andover: Cengage Learning EMEA.
Hauck-Filho, N., and Valentini, F. (2020). Considerações teóricas e práticas sobre a confiabilidade dos instrumentos psicológicos: um tutorial. Psico-USF 25, 249–259.
Henseler, J., Ringle, C. M., and Sarstedt, M. (2015). A new criterion for assessing discriminant validity in variance-based structural equation modeling. J. Acad. Market. Sci. 43, 115–135. doi: 10.1007/s11747-014-0403-8
Hore-Lacy, F., Gwini, S., Glass, D. C., Dimitriadis, C., Jimenez-Martin, J., Hoy, R. F., et al. (2024). Psychometric properties of the Perceived Stress Scale (PSS-10) in silica-exposed workers from diverse cultural and linguistic backgrounds. BMC Psychiatry 24:181. doi: 10.1186/s12888-024-05613-6
Igra (2024). A systematic – review of academic stress intended to improve the educational journey of learners. Methods Psychol. 11:100163. doi: 10.1016/j.metip.2024.100163
International Test Commission (2017). The ITC Guidelines for Translating and Adapting Testes, 2nd Edn. Instituto Brasileiro de Avaliação Psicológica (Trad.). (Accessed March 15, 2025).
JASP Team (2025). JASP (Version 0.19.3) [Computer Software]. Available online at: https://jasp-stats.org/download/ (Accessed March 20, 2025).
Jatic, Z., Trifunovic, N., Erkocevic, H., Hasanovic, E., Dzambo, I., and Pilav, A. (2023). Construct validity of the Perceived Stress Scale (PSS-10) in a sample of health professionals in family medicine in Bosnia and Herzegovina. Public Health Pract. 6:100413. doi: 10.1016/j.puhip.2023.100413
Kogar, E. Y., and Kogar, H. (2024). A systematic review and meta-analytic confirmatory factor analysis of the Perceived Stress Scale (PSS-10 and PSS-14). Stress Health 40:e3285. doi: 10.1002/smi.3285
Li, C. H. (2016). Confirmatory factor analysis with ordinal data: comparing robust maximum likelihood and diagonally weighted least squares. Behav. Res. Methods 48, 936–949. doi: 10.3758/s13428-015-0619-7
Liu, X., Zha, Y., Li, J., Dai, J., Wang, X., and Wang, S. (2020). Factor structure of the 10-item Perceived Stress Scale and measurement invariance across genders among Chinese adolescents. Front. Psychol. 11:537. doi: 10.3389/fpsyg.2020.00537
Luft, C. D. B., Sanches, S. O., Mazo, G. Z., and Andrade, A. (2007). Versão brasileira da Escala de Estresse Percebido: Tradução e validação para idosos. Rev. Saúde Pública 41, 606–615. doi: 10.1590/S0034-89102007000400015
Machado, W. L., Damásio, B. F., Borsa, J. C., and da Silva, J. P. (2014). Dimensionalidade da Escala de Estresse Percebido (Perceived Stress Scale, PSS-10) em uma Amostra de Professores. Psicol. Reflexão Crítica 27, 38–43. doi: 10.1590/S0102-79722014000100005
Makhubela, M. (2022). Assessing psychological stress in South African university students: measurement validity of the Perceived Stress Scale (PSS-10) in diverse populations. Curr. Psychol. 41, 2802–2809. doi: 10.1007/s12144-020-00784-3
Marakshina, J., Adamovich, T., Vasin, G., Ismatullina, V., Lobaskova, M., Malykh, A., et al. (2024). Factor structure and psychometric properties of the Perceived Stress Scale in Russian adolescents. Sci. Rep. 14:775. doi: 10.1038/s41598-023-51104-1
Mardia, K. V. (1970). Measures of multivariate skewness and kurtosis with applications. Biometrika 57, 519–530. doi: 10.1093/biomet/57.3.519
Markon, K. E. (2019). Bifactor and hierarchical models: specification, inference, and interpretation. Annu. Rev. Clin. Psychol. 15, 51–69. doi: 10.1146/annurev-clinpsy-050718-095522
Marôco, J. (2014). Análise de equações estruturais: Fundamentos te?ricos, software & aplicações. Pêro Pinheiro.
Mattos, L. A. P., de, Lima, M. A. G., Moura, T. C. C., and Bastos, F. M. C. (2020). Avaliação da saúde mental de estudantes universitários em tempos de pandemia: um estudo com alunos de uma instituição pública de ensino superior. Rev. Brasil. Psicol. 7, 18–29.
Mendis, B. I. L. M., Palihaderu, P. A. D. S., Karunanayake, P., Satharasinghe, D. A., Premarathne, J. M. K. J. K., Dias, W. K. R. R., et al. (2023). Validity and reliability of the Sinhalese version of the Perceived Stress Scale questionnaire among Sri Lankans. Front. Psychol. 14:1152002. doi: 10.3389/fpsyg.2023.1152002
Mondo, M., Sechi, C., and Cabras, C. (2021). Psychometric evaluation of three versions of the Italian Perceived Stress Scale. Curr. Psychol. 40, 1884–1892. doi: 10.1007/s12144-019-0132-8
Murakami, K., Santos, J. L. F., dos, Troncon, L. E., de, A., and Panúncio-Pinto, M. P. (2024). Estresse e Enfrentamento das Dificuldades em Universitários. Psicol. Ciência Profissão 44, e258748, 1–16.
Nunes, C. H. S. S., and Primi, R. (2010). “Aspectos técnicos e conceituais da ficha de avaliação dos testes psicológicos,” in Avaliação psicológica: diretrizes na regulamentação da profissão (Campinas: Conselho Federal de Psicologia), 101–128.
Perera, M. J., Brintz, C. E., Birnbaum-Weitzman, O., Penedo, F. J., Gallo, L. C., Gonzalez, P., et al. (2017). Factor structure of the Perceived Stress Scale-10 (PSS) across English and Spanish language responders in the HCHS/SOL Sociocultural Ancillary Study. Psychol. Assess. 29, 320–328. doi: 10.1037/pas0000336
Reis, D., Lehr, D., Heber, E., and Ebert, D. D. (2017). The German version of the Perceived Stress Scale (PSS-10): evaluation of dimensionality, validity, and measurement invariance with exploratory and confirmatory bifactor modeling. Assessment 26, 1246–1259. doi: 10.1177/1073191117715731
Reis, R. S., Hino, A. A. F., and Añez, C. R. R. (2010). Perceived stress scale: reliability and validity study in Brazil. J. Health Psychol. 15, 107–114. doi: 10.1177/1359105309346343
Reise, S. P., Bonifay, W. E., and Haviland, M. G. (2018). “Bifactor modelling and the evaluation of scale scores,” in The Wiley Handbook of Psychometric Testing: A Mulitidiciplinary Reference on Surevey, Scale and Test Development, Vol. 2, eds. P. Irwing, T. Booth, and D. J. Hughes (New York, NY: John Wiley and Sons), 677–708. doi: 10.1002/9781118489772.ch22
Ribeiro, S., Teles, H., Ramalho, N., and Ramalho, V. (2024). Perceived Stress Scale (PSS-10) Aplicada a Assistentes Sociais em Portugal: Estudo das Propriedades Psicométricas. Euro. J. Soc. Sci. Stud. 10, 116–128. doi: 10.46827/ejsss.v10i2.1769
Rodriguez, A., Reise, S. P., and Haviland, M. G. (2015). Applying bifactor statistical indices in the evaluation of psychological measures. J. Pers. Assess. 98, 223–237. doi: 10.1080/00223891.2015.1089249
Ruisoto, P., López-Guerra, V. M., Paladines, M. B., Vaca, S. L., and Cacho, R. (2020). Psychometric properties of the three versions of the Perceived Stress Scale in Ecuador. Physiol. Behav. 224:113045. doi: 10.1016/j.physbeh.2020.113045
Seabra-Santos, M. J., Silva, C. F., Pereira, A. I., Ribeiro, L., Lima, V., and Gaspar, M. F. (2021). Avaliação em Psicologia Clínica e da Saúde: Estudos com adultos e idosos. Coimbra: Imprensa da Universidade de Coimbra.
She, Z., Li, D., Zhang, W., Zhou, N., Xi, J., and Ju, K. (2021). Three versions of the Perceived Stress Scale: psychometric evaluation in a nationally representative sample of Chinese adults during the COVID-19 pandemic. Int. J. Environ. Res. Public Health 18:8312. doi: 10.3390/ijerph18168312
Souza, Á., Silva, F. L., da Dinelly, M. M., Botelho, S. O., Lorenzoni, S. M., de, A., et al. (2025). Formação de professores de ciências naturais na região norte do Brasil: Uma análise do período de 2018 a 2024. Rev. Aracê 7, 5708–5715. doi: 10.56238/arev7n2-069
Tabachnick, B. G., and Fidell, L. S. (2019). Using Multivariate Statistics, 7th Edn. Boston, MA: Pearson.
Trigo, M., Canudo, N., Branco, F., and Silva, D. (2010). Estudo das propriedades psicométricas da Perceived Stress Scale (PSS) na população portuguesa. Psychologica 53, 353–378. doi: 10.14195/1647-8606_53_17
World Health Organization (2023). Stress. Available online at: https://www.who.int/news-room/questions-and-answers/item/stress (Accessed February 10, 2025).
Xiao, T., Zhu, F., Wang, D., Liu, X., Xi, S.-J., and Yu, Y. (2023). Psychometric validation of the Perceived Stress Scale (PSS-10) among family caregivers of people with schizophrenia in China. BMJ Open 13:e076372. doi: 10.1136/bmjopen-2023-076372
Keywords: stress, anxiety, psychometric properties, Brazilian university students, PSS-10
Citation: Ferreira LS, Silva SSdCe, Pontes FAR, Hernandez JAE, Silva EP, Queiroz LC, Käppler CdO and Silva JCe (2026) Psychometric evaluation of the Perceived Stress Scale among university students in Northern Brazil. Front. Educ. 10:1695518. doi: 10.3389/feduc.2025.1695518
Received: 29 August 2025; Revised: 15 November 2025;
Accepted: 28 November 2025; Published: 16 January 2026.
Edited by:
Vitor Franco, University of Evora, PortugalReviewed by:
Juan P. Sanabria-Mazo, Universitat Rovira i Virgili, SpainLaís Santos Vitti, Pontifical Catholic University of Campinas, Brazil
Copyright © 2026 Ferreira, Silva, Pontes, Hernandez, Silva, Queiroz, Käppler and Silva. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Laiana Soeiro Ferreira, bGFpYW5hLnNvZWlyb0B1ZnBhLmJy