INTRODUCTION

In recent decades, resilience has been one of the constructs that has aroused the most interest in the clinical and educational fields, being a field of knowledge where observations, research and psychosocial practices converge, demonstrating the ability of people to resist and overcome adverse situations and to build themselves in an integral way, despite having lived through such experiences (^{Uriarte, 2005}).

^{Smith et al. (2008}) state that resilience is the ability to adapt and recover from stress or illness and to thrive in the face of adversity, therefore, it is understood that although individuals face unpleasant situations that generally turn into problems, a resilient person sees these difficulties as a challenge to which he or she must adapt using personal resources.

^{Smith et al. (2010}) build on the theory proposed by ^{Carver (1998}), who suggests that there are at least four possible outcomes when a person faces adversity. The first possibility is a response in which the initial adverse effect is exacerbated and the individual eventually succumbs, another outcome is that the person survives but is diminished or impaired in some way. A third outcome is a rapid or gradual return to the level of functioning prior to facing adversity, and the last possibility is that the person does not simply return to the previous level, but can cope with it. For this author, the assessment of the possibility that a person can get better after adversity is what is referred to by the term resilience, which is the ability to recover and thrive, that is, to move to a higher level of functioning after going through a stressful event, which denotes the return to a previous state of homeostasis.

Resilience includes the ability to build appropriate interpersonal relationships, work, enjoyment, and psychological well-being in the face of inevitable problems. Therefore, ^{Uriarte (2005}) argues that being resilient means that the person has adequate mental health. Likewise, ^{Scheier and Carver (1985}) mention that resilience is related to optimism, as the latter plays a key role in adaptive coping behaviors. They also report that optimistic people behave differently than pessimistic people.

In addition, there are risk factors that affect an individual’s well-being, such as anxiety, stress, and depression, which are more prevalent among university students (^{Gutiérrez et al., 2010}). A study by ^{Fínez and Morán (2015}) shows that there is an inverse relationship between resilience and anxiety, because if a student is able to resolve the setbacks that occur throughout his or her life, he or she will be less anxious in the face of the demands that he or she may face.

Depression is considered to be a factor related to the perception of quality of life, which may be accentuated in adolescence and may persist to a greater degree in later stages, affecting the individual’s functionality (^{Pardo et al., 2004}). Therefore, resilience is considered a protective factor that allows for a better quality of life and promotes personal well-being. Likewise, it is believed that for the resilience process to occur, the person must be exposed to a level of stress that affects his or her homeostasis and through which the individual can adapt positively (^{Monroy & Palacios, 2011}). In this regard, ^{García et al. (2016}) mention that there is an interaction between resilience and health, in this way, people with low resilience tend to have a late recovery in the face of adversity, which predisposes the appearance of risk factors related to coping with an illness, either physical or psychological.

As a result, various resilience measurement instruments have been developed over the years, due in part to the wide range of definitions of the construct (^{Cantero & Alonso, 2018}). Some of these instruments measure resilience directly, while others do so indirectly the various factors that make up resilience (^{Ortunio & Guevara, 2016}). Among the most widely used instruments at the international level is the CD-RISC Scale by ^{Connor and Davidson (2003}), which even has short versions in Spanish. However, as Cantero and Alonso (2018) mentioned, while these instruments measure factors associated with resilience, they do not assess resilience as such, as they do not consider the elements of adversity and positive adaptation proposed by ^{Luthar et al. (2000}).

In this regard, ^{Windle et al. (2011}) argue that, in terms of the instruments reviewed, the Brief Resilience Scale (BRS) developed by ^{Smith et al. (2008}) assesses these elements and therefore measures resilience in its most basic sense and is considered by this author to be one of the resilience instruments with the highest quality in its measurements when compared to 19 other instruments. To the above, it should be added that, in its Spanish version translated by ^{Rodríguez et al. (2016}), it is a short and easy-to-understand instrument. Likewise, the BRS has been the subject of an analysis of its psychometric properties at the international level, with positive results. These studies include Rodriguez et al. (2016) among 620 Spanish adults, including parents of children with health problems, individuals with health problems, and the general population; ^{Kyriazos et al. (2018}) among Greek adults from the general population; ^{Fung (2020}) among 511 Chinese university students; and ^{Peña-Contreras et al. (2020}) among 648 Ecuadorian adults from the general population.

In relation to the above, an analysis of the psychometric properties of validity, reliability and fairness of the BRS in Peruvian university students should be considered as an objective, aiming to provide empirical support to the construct of resilience from the author’s conception of the instrument, as a measurement that takes into account the components of adversity and positive adaptation. It will also make it possible to apply the BRS correctly and provide new psychometric evidence for using the BRS in future research.

METHOD

RESULTS

Validity evidence based on internal structure

For the confirmatory factor analysis, the WLSMV estimation method was used, proving that the oblique model (M3) is the one that presents better fit indices compared to the other models presented; these indices were: CFI= .99; TLI= .98; RMSEA= .036 and SRMR= .023, as shown in Table 2.

Table 1 Descriptive analysis of the BRS items (n=468) 

Note: M= Mean; SD= Standard deviation; g¹= skewness; g²= kurtosis; CHI= corrected homogeneity index; h²= Communality; DI= Discrimination index by comparison of extreme groups.

Source. Elaborated by the author.

Table 2 Fit indices for BRS factor structure models 

Note: X²= Chi-square; gl= Degrees of freedom; X²/gl= Chi-Square / Degrees of Freedom Ratio; CFI= comparative fit index; TLI= Tucker-Lewis index; RMSEA= Root mean squared error and SRMR= Standardized response mean.

Source. Elaborated by the author.

Validity evidence in relation to other variables

Table 3 shows the correlation analysis for the criterion validity and convergence of the BRS in relation to the BRCS and the LOT-R, with Pearson correlation coefficients of .606 and .617, respectively. A direct relationship was found, meaning that the higher the level of resilience, the more optimistic the person tends to be. Similarly, criterion validity is provided between two scales measuring the resilience construct. Moreover, for the BRS and DASS-21 scales, coefficients of -.546 with depression, -.515 with anxiety, and -.514 with stress were found, indicating that there is an inversely proportional relationship between resilience and depression, anxiety, and stress.

Table 3 Relationship between resilience, optimism, depression, anxiety and stress 

Note: r= Pearson correlation coefficient; p= significance level

Source. Elaborated by the author.

Equity

The equity assessment was performed by factorial invariance, as shown in Table 4. Four levels of invariance were assessed: configural, factorial loads, intercepts and residuals, considering CFI and RMSEA fit indices and their differences between each level. To confirm the presence of invariance, the values Δ CFI <.01 and Δ RMSEA <.015 must be found (^{Chen, 2007}; ^{Cheung & Rensvold, 2002}). The aforementioned criteria are met in the four levels of invariance assessed, in terms of gender, age and university. Therefore, as a whole, it can be affirmed that the scale is invariant according to gender, age and university, thus showing equity in its measurements.

Table 4 BRS equity analysis 

Note: Δ CFI = variation in CFI and Δ RMSEA = variation in RMSEA.

Source. Elaborated by the author.

DISCUSSION

This research aimed to analyze the psychometric properties of the Brief Resilience Scale (BRS) in a sample of Peruvian university students. This instrument is a brief scale that assesses resilience to stress in the face of adverse situations (^{Smith et al., 2008}).

The analysis of the items was carried out, obtaining acceptable values in the homogeneity index, which indicates that the items tend to measure the same construct. As for the communalities, values >.40 were observed, indicating that the items are related (^{Lloret et al, 2014}). However, in item 2 the communality is .25 which is considered a minimum condition because it is less than .30 (^{Conway & Huffcutt, 2003}). Also, the discrimination index by the extreme group comparison method is statistically significant p<.001, this shows that the items have the ability to differentiate greater or lesser presence of resilience. (^{Cohen & Swerdlik, 2002}).

Subsequently, the factorial structures were compared using different models, showing that the oblique model had better fit indices compared to the unidimensional and orthogonal models. This is similar to the result found by ^{Fung (2020}), in which the two-factor model proved to have better fit indices compared to the unidimensional model. Similar results were obtained by ^{Peña-Contreras et al. (2020}). Although the BRS was originally proposed as a unidimensional scale with negative items, it should be noted that this scale was developed only on the basis of the results obtained from the analysis of the principal components. However, subsequent research using confirmatory factor analysis to confirm the factor structure suggests that the instrument has two latent factors, the first based on the positive items related to resilience and the second based on the negative items related to succumbing. Also, in the context of confirmatory factor analysis, both ^{Rodriguez et al. (2016}) and ^{Kyriazos et al. (2018}) have pointed out that although reverse item wording serves to avoid acquiescence bias, it creates an effect that contributes to the formation of two factors. It is therefore suggested that this effect be taken into account in the analysis of models relating to this instrument.

Reliability was assessed using alpha and omega coefficients and acceptable values were obtained (α=.799, ω=.853). Although it is true that one of the most widely used methods to determine reliability is through the alpha coefficient, it should be considered that this has limitations, so it is convenient to use another coefficient such as the omega coefficient (^{Ventura & Caycho, 2017}). These results are consistent with those of the authors mentioned above.

For convergent and divergent criterion validity, the Pearson correlation coefficient was used to determine the degree of relationship between the BRS and other variables. The following instruments were used: Brief Resilient Coping Scale BRCS, Life Orientation Test LOT -R and Depression Anxiety Stress Scale-21 DASS-21. In terms of criterion validity, this occurs when instruments measure the same construct (^{Argibay, 2006}). Therefore, a correlation was made with the BRCS, which was found to be a brief scale with adequate psychometric properties in older Peruvian adults (^{Caycho et al., 2018}). A Pearson correlation coefficient of r =.606 was obtained, indicating a high degree of relationship (^{Cohen, 1988}).

In terms of convergent validity, the BRS obtained positive and significant correlations with the LOT-R (^{Fung, 2020}; ^{Kunzler et al., 2018}) (r=.617) and with the BRCS (r=.606). In contrast, regarding divergent validity, the correlations with DASSS-21 were r=-.546 for depression, r=-.515 for anxiety, and r=-.514 for stress. ^{Kyriazos et al. (2018}), in their study, correlated BRS and DASS -21 obtaining the strongest value with stress, while in this study it was with depression. In all cases, the coefficients were considered high (^{Cohen, 1988}).

Regarding factorial invariance analysis, the results show that the structure of the two-factor correlated model is equivalent regarding gender, age, and university due to goodness of fit to the sample (n= 468). These results indicate that the scale and its items have the same meaning for male and female participants from different age groups and from different universities, whether public or private. Therefore, the scale items measure the latent variable in a similar way in the different groups (^{Brown, 2015}). In light of this, it can be affirmed that if differences are found in the scores according to the groups mentioned, they would be due to the greater or lesser presence of resilience and not to measurement bias (^{Domínguez, 2016}). The same result regarding gender invariance was found by ^{Peña-Contreras (2020}) in Ecuadorian adults.

In conclusion, the results obtained specify that the BRS has psychometric evidence of internal structure, criterion, convergent and divergent validity, as well as internal consistency and fairness reliability, which makes it an ideal instrument for its application in Peruvian university students. However, the content validity remains to be assessed, especially in the case of item 2, for which the communality value was lower than expected. Finally, a review of the psychometric properties of the scale in different samples is recommended to provide more evidence for its use