BACKGROUND

Currently, there are highly stressful events that generate mental health problems in the world and national population, related to a high incidence of stress, anxiety, depression, decreased optimism (^{Santos et al., 2022}), negative psychological impact (^{Brooks et al., 2020}) and overflow in the capacity to respond to stress (^{Guillén-Díaz-Barriga et al., 2021}).

In complex situations of such magnitude, coping with stress provides evidence of how people act to a stressful event. Here it can be highlighted that some coping strategies cushion the negative effect of stressors, while others lead to avoiding them; but all this depends on the perceived control of the event (^{Dijkstra & Homan, 2016}).

In a more functional sense, coping with stress acts as a stabilizing factor because it facilitates adequate personal adjustment to a stressful situation (^{Morán et al., 2010}), also allows the adaptation of cognitive schemes and regulates the perception of the stressor threat that alters the adequate coping (^{Mate et al., 2016}).

From a cognitive approach, coping with stress refers to the efforts made by the person to manage the demands of adaptation when interacting with their context (^{Lazarus, 2006}) and tries to prevent or reduce the threats, damage and anguish generated by stress (^{Carver & Connor-Smith, 2010}). In this sense, stress arises from the person-context interaction, in addition to being classified as changing all the time. Now, when these experiences of stress arise, the person performs a primary evaluation where a set of values, beliefs and cognitive filters influence; while in the secondary evaluation he makes available his personal resources to face the event; both forms of evaluation determine the actual damage or loss generated by stress (^{Folkman, 2010}).

These two evaluative forms are synthesized in responses to stress focused on the problem and emotion (^{Lazarus & Folkman, 1984}). In the first, the object of study is in the planned resolution of the problem, while in the second it is intended to regulate the emotion with negative impact, using evasive strategies such as distancing, the search for emotional support and flight-avoidance (^{Folkman, 2010}).

While some forms of stress response have allowed adaptive coping towards stressors; others, on the other hand, have received criticism in the scientific literature because they have always associated negative emotions as those that drive coping; however, to correct these shortcomings, the new conceptions develop a third style focused on meaning and positive emotions, which help restore and maintain problem- and emotion-centred coping in the long term (^{Folkman, 2008}).

On the other hand, different positions have emerged that relate coping as something situational (^{Lazarus, 2006}). This theoretical position served to design different measuring instruments; however, the psychometric inconsistencies reported in initial studies led to the inclusion of dispositional coping to give greater consistency to the construct (^{Carver et al., 1989}).

Thus, the COPE 28 Brief scale measures stress coping in its situational or dispositional version, maintaining its orientation of state and trait (^{Carver & Connor-Smith, 2010}) In this regard, several studies that used the scale reported significant psychometric inconsistencies, since the evidence of validity throws several factors and low reliability (^{Solberg et al., 2022}).

In the Spanish version, the critical point was located in the reliability of the factors, but that after the second-order factor analysis, 4 factors were found that explain 52.9% of the variance and an improvement in reliability 0.71 to 0.80 (^{Morán et al., 2010}); however, in a Portuguese sample, the factors reported were 14, similar to the original structure, whose goodness indices are appropriate and factorial loads greater than 0.40 (^{Nunes et al., 2021}).

Studies using populations with health problems, cultural characteristics or evolutionary stages also found discrepancies in internal structure. For example, in French patients and caregivers, 4 factors were evidenced with adequate adjustment of goodness RMSEA=0.047; CFI=0.923 and RMSEA=0.031; CFI=0.938 (^{Baumstarck et al., 2017}); in people with liver transplantation, 6 factors were found (^{Amoyal et al., 2016}); in breast cancer survivors, the 14-factor model showed the best fit (^{Rand et al., 2019}) and in (^{Amoyal et al., 2016}; ^{Tang et al., 2021}).

On the other hand, in people with aggressiveness, the 4-factor model yielded adequate adjustments of kindness (^{Cramer et al., 2020}); while in other cultures a three-factor model was found with adequate adjustment indices GFI = 0.924; TLI=0.904; RMSEA=0.039 and good reliability 0.84, 0.75 and 0.81 (^{Alghamdi, 2020}).

Regarding the Latin American context, a study with Argentine older adults reported that the model with the best fit was two factors CFI = 0.937; TLI=0.908; RMSEA=0.091) and a reliability of 0.81 (^{Richard’s et al., 2021}). On the contrary, Chilean adaptation found structure of 14 factors with expected goodness indices and good internal consistency (^{García et al., 2018}). As for Peru, no psychometric studies of COPE 28 have been reported, but in the version of 52 items itself (^{Cassaretto Bardales & Chau Perez-Aranibar, 2016}).

As can be seen in various studies, the scale has different factorial structures. This is due to the limitations of methodological nature, as well as practical functionality such as, for example, having been adapted exclusively in specific samples and with university students; also, to the validations in their dispositional version, use it in a dichotomous response modality, have a disparate number of items in each scale and for presenting some items with negative charge (^{Cano et al., 2007}; ^{Guillén-Díaz-Barriga et al., 2021})

These limitations and the gaps in studies in the Peruvian context led us to analyze the psychometric evidence of the COPE 28 scale in a Peruvian population. In this sense, the study is justified because it provided relevant background on the internal structure of the instrument, so as to strengthen the theory of coping with stress and confirm factors of the scale in the Peruvian context. It also provides an instrument with psychometric properties of validity and reliability for other researchers to use in different contexts of our country.

The objective was to determine the psychometric evidence of reliability and validity of the BRIEF COPE 28 scale in a Peruvian population.

METHODS

RESULTS

In a first analysis, it can be verified that the item-test correlations range between 0.223 and 0.508; p<0.05, considered acceptable. Likewise, the internal consistency of the items yields an alpha coefficient higher than α=0.80, indicating good reliability (^{Furr, 2011}; ^{Reidl-Martínez, 2013}) (Table 1).

Table 1 Analysis of the internal consistency of COPE items 28 

Note. M=Mean, SD=standard deviation; r=correlation α=Alpha coefficient.

Regarding the internal consistency of the COPE 28 BRIEF by dimensions, it is observed that, in coping styles, the ordinal alpha coefficient of the cognitive style is α ordinal = 0.82, social support α ordinal = 0.81, blocking coping α ordinal = 0.74 and spiritual coping α ordinal =0.75. In the same line, the internal consistency of coping strategies shows that the lowest value is found in Self-distraction α ordinal = 0.58, while the highest value in the use of substances α ordinal = 0.93. These internal consistency values are acceptable as they meet the established criteria and the instrument is reliable (^{Furr, 2011}; ^{Reidl-Martínez, 2013}). It can also be evidenced that the correlations of coping styles yield values above r=0.440, and in coping strategies higher than r=0.234, which are statistically significant p<0.05 (Table 2).

Table 2 Reliability of COPE 28 coping styles and strategies 

Note. α= alpha coefficient; ordinal α= ordinal alpha coefficient

The first model retains the 4 styles of the original version adapted to Spanish, whose adjustment indexes do not comply with the established (Table 3). For this reason, items 3, 4, 5, 7, 8, 11, 12, 13, 19, 21 and 26 (Table 4) with low factor loads <0.30 were eliminated and a second model was tested, which retains the 4 styles, but with 17 items. It was found that the goodness adjustment indices SB-χ²/gl=1.836; CFI=0.92; TLI=0.90 are within the established to be considered a good fit of the model; however, SRMS=0.09 and RMSEA=0.10 have values higher than allowed (^{Brown, 2015}); these indicators are influenced by the amount of the sample, but demonstrate that the tested model is inconclusive (Table 3).

On the other hand, in the evidence of validity of coping strategies, 4 models were tested. The first model was performed with the 14 factors of the original version, which showed adjustment rates below the established, which led to the purification of the strategies of acceptance, denial, self-incrimination and relief (Table 5) that had factorial loads <0.30. Subsequently, three other models were tested, which show good goodness adjustments; however, the fourth model consists of 10 strategies with adequate fit indices SB-χ²/gl=1.902; CFI=0.96; TLI=0.95, SRMS=0.056 and RMSEA=0.069 and meet the statistical criteria (^{Brown, 2015}) to determine that BRIEF COPE 28 presents acceptable psychometric properties (Table 3).

Table 3 Confirmatory factor analysis of the coping styles and strategies of COPE 28 

Note. Relative Chi-square (2/gl); comparative adjustment index (CFI); Tucker-Lewis index (TLI); Stan-dardized Residual Mean Square Root (SRMR); Mean square error of approximation (RMSEA).

Table 4 Items eliminated in the analysis of the four styles. 

Note: Model with 4 styles and 17 items.

Table 5 Strategies that make up each of the models of confirmatory factor analysis. 

Note: The four models with their coping strategies.

DISCUSION

The objective of this study was to determine the psychometric evidence of reliability and validity of BRIEF COPE 28, Spanish version (^{Morán et al., 2010}) in a Peruvian population. The results of the internal consistency analysis show item-test correlations superior to 0.223 and acceptable reliability values in cognitive style α ordinal = 0.82, social support α ordinal = 0.81, blocking coping α ordinal = 0.74 and spiritual α ordinal = 0.75. Likewise, it was reported that in 8 coping strategies the ordinal alpha ranges from 0.72 to 0.93 (^{Furr, 2011}; ^{Reidl-Martínez, 2013}), while the lowest values were evidenced in self-distraction and planning.

These results are similar to studies that reported alpha coefficients greater than 0.80 for coping styles (^{Cramer et al., 2020}). Regarding coping strategies, some studies reported similar reliability values between 0.71 and 0.82 (^{Baumstarck et al., 2017}), 0.64 (^{Amoyal et al., 2016}), 0.57 (^{Baumstarck et al., 2017})) and 0.93 (^{Tang et al., 2021}). It is also consistent with low scores for distraction (α=0.43) and high scores for substance use (α=0.88) (^{Nunes et al., 2021}). On the other hand, similarities were found in item-test correlations greater than 0.20 (Amoyal et al., 2016), between 0.20 and 0.63 (^{Tang et al., 2021}) and between 0.30 and 0.80 (^{Morán et al., 2010}).

Regarding the evidence of validity, the model with the greatest adjustment of goodness was that of 17 items with the four coping styles, whose indices of χ²=1.836; CFI=0.92; TLI=0.90 were adequate; however, the parsimony adjustments SRMS=0.09 and RMSEA=0.10 present values higher than the established statistical criterion (^{Brown, 2015}). This indicates that the model of 17 items only has adjustment in relative Chi-square and in the comparative indices, which coincide with what was found in systematic studies, since it reaffirms the inconsistency of the internal structure, especially in the number of factors extracted and the fit of the models (^{Solberg et al., 2022}).

On the other hand, there are studies that have also reported the presence of four factors with good adjustment in the RMSEA = 0.047 and CFI = 0.923 (^{Baumstarck et al., 2017}); the same was found in an exploratory factor analysis, whose four factors explain 52% of the variance (^{Morán et al., 2010}). However, it differs with studies that reported a three-factor structure GFI=0.924; TLI=0.904; RMSEA=0.039 (^{Alghamdi, 2020}); similarly, it disagrees with the four-factor adjustment indices CFI=0.089; RMSEA=0.108; SMRM=0.078 reported in aggressive people (^{Cramer et al., 2020}), and with all three factors found in a sample of Chinese children (^{Tang et al., 2021}).

Regarding coping strategies, it was found that the fourth model consisting of the 10 strategies presents a good fit χ²=1,902; CFI=0.96; TLI=0.95; SRMS=0.056, RMSEA=0.069 (^{Brown, 2015}) but it is also found that there is a good fit in the second and third models with 13 and 12 strategies (see Table 3). The results found differ with the internal structure of 14 strategies of the Portuguese sample χ2=1.06; CFI= 0.99; RMSEA = 0.02 (^{Nunes et al., 2021}), with the Chilean adaptation (^{García et al., 2018}) and in the sample of breast cancer survivors (^{Rand et al., 2019}). It also differs with the internal structure of 8 factors CFI = 0.937; TLI=0.908; RMSEA=0.091 (^{Richard’s et al., 2021}) and with the internal structure of 6 factors (^{Amoyal et al., 2016}).

With what it is reported in this research and other cited studies, it is found that the internal structure of COPE 28 has important variations in terms of the original (^{Carver et al., 1989})1989); these variations are explained due to the context and characteristics of the studied samples (^{Solberg et al., 2022}) but it should be noted that the four coping styles maintain its structure, reflecting the theoretical consistency of the instrument.

On the other hand, the methodological implications extracted allow evaluating the instrument from a dispositional and situational perspective to improve the internal structure of COPE 28 (^{Carver & Connor-Smith, 2010}) since coping styles present less adjustment and require more robust methodologies. However, in general terms, the results obtained from COPE 28 show adequate psychometric properties that can be contrasted with other studies.

Limitations

Regarding limitations, the selection of the sample was carried out with a non-probabilistic method, which interferes to generalize results; in addition, the sample size had an important influence on the fit of the reported models. On the other hand, the collection technique used is prone to biases that can influence the results. For this reason, other studies suggest increasing the sample size and better characterizing it to achieve better heterogeneity. It is also suggested to use the instrument in different contexts and populations to further improve internal consistency.

CONCLUSION

In conclusion, the Spanish version of BRIEF Cope 28 presents adequate evidence of validity and reliability. Regarding coping styles, the four-style model obtains good comparative fit, but not in the parsimony index, so the results are not yet conclusive. On the other hand, the model of 10 factors or coping strategies has optimal adjustment rates, which allows us to affirm that COPE 28 has adequate validity.