Introduction

From a cognitive perspective, ^{Lazarus and Folkman (1984}a) defined coping as a set of efforts made to change the environment in an attempt to adapt in the best possible way to a stressful event, reducing or minimizing its aversive character. Changes can be cognitive or behavioral and are meant to help the individual manage situations that exceed their personal resources. For the authors, coping strategies can be learned and maintained or not throughout life, depending on the history of each individual, and can be problem-focused or emotion-focused.

For ^{Folkman and Lazarus (1980}), problem-focused coping comprises efforts to identify the problem, determine solutions and alternatives, assess the costs and benefits of actions, adopt attitudes to change what is possible and, if necessary, learn new skills related to the desired or expected result. When focusing on the problem, individuals seek to control the stressor and actions are aimed at reducing or eliminating it, which are considered to be more resolving strategies.

On the other hand, emotion-focused coping strategies are usually used in situations identified as unchangeable, being more palliative. In this type of strategy, the individual's emotion is modulated when facing a stressful situation, in an attempt to reduce the unpleasant sensation caused by stress. An example of that is a situation in which a person prays and feels better by doing so. However, it is important to remember that coping strategies are interrelated, as different people might analyze a stressful event differently. For one individual, the strategy used to solve the problem can turn to emotion, while for another it can focus on the problem, or both strategies can be used at the same time. A student can study to pass a test (problem) and also pray to be able to do well (emotion) (^{Folkman, & Lazarus, 1980}).

Furthermore, the authors have developed and described a coping assessment model using these two forms (problem and emotion) and created the Ways of Coping Checklist (WCC), a clinical measure to assess coping. The WCC assesses stress coping strategies and experiences in a clinical context using 67 items, which can be responded as "yes" or "no". Subsequently, ^{Folkman and Lazarus (1988}) expanded their investigations about the instrument by reviewing data that were initially based on the literature. Through an empirical strategy, they proposed a new instrument, the Ways of Coping Questionnaire (WCQ). The main change was related to the form of the answers, which went from a checklist to a Likert scale. The WCQ consisted of 66 items from the Ways of Coping Checklist, encompassing thoughts, actions and strategies used by individuals to deal with internal and external demands of a specific stressful event, and ordinal responses with four possibilities. Individuals were expected to respond indicating the frequency in which they used each strategy, according to the following order, 0- not used; 1- used somewhat; 2- used quite a bit; 3- used a great deal. Items were selected in a qualitative way, and those considered redundant or unclear by the authors were removed or revised, and several items were changed at the suggestion of respondents (^{Folkman & Lazarus, 1988}).

Based on the original study and some psychometric studies of the Ways of Coping Questionnaire, the instrument was subdivided into eight coping strategies: Confrontation, Distancing, Self-control, Social Support, Accepting Responsibility, Escape-Avoidance, Problem Solving, and Positive Reappraisal (^{Coyne, Aldwin, & Lazarus, 1981}; ^{Folkman & Lazarus, 1985}; ^{Folkman, Lazarus, Gruen, & Delongis, 1986}; ^{Folkman & Lazarus, 1988}). Internal consistency analysis for the eight factors proposed by ^{Lazarus and Folkman (1988)} indicated reliability coefficients between .61 (distancing) and .79 (positive reappraisal).

Other studies were conducted to look for validity evidence for the Ways of Coping Questionnaire.^{Vitaliano et al. (1985}) found a five-factor solution with 42 items in the United States. ^{Bramsen et al. (1995}), in turn, found seven factors when adapting the Ways of Coping Questionnaire, to the Netherlands. In Norway, ^{Falkum, Oiff and Aasland (1997}) identified a six-factor structure, maintaining the 42 items of Vitaliano et al. (1985). More recently, ^{Liew, Santoro, Edwards, Kang and Cronan (2016}) adapted the instrument to the healthcare context, finding a four-factor structure for patients from the United States with fibromyalgia, while ^{Corti et al. (2018}) found a six-factor structure for Australian Parkinson's patients. These differences between the factor structures found are expected, as they are part of an adaptation of the instrument to different countries and cultures (^{Borsa & Seize, 2017}).

Despite the different factor structures found, the Ways of Coping Questionnaire is considered one of the most widely used instruments to measure coping strategies (^{Hirsh et al., 2015}; ^{Pais-Ribeiro & Santos, 2001}). In Brazil, the instrument was translated and adapted by ^{Savóia, Santana and Mejias (1996}). The latter authors, in their translation and adaptation study, named the instrument Inventário de Estratégias de Coping - IEC (Coping Strategies Inventory - CSI).

In Brazil, the CSI was translated and adapted from the original version with 66 items and eight factors developed by ^{Folkman and Lazarus (1988}). ^{Savóia et al. (1996}) verified the adequacy of the CSI translation into Brazilian Portuguese, as well as its validity evidence based on internal structure and relationship with other variables. First, the instrument was translated into Portuguese and analyzed by judges. Results indicated that, despite cultural and semantic issues, the translation remained true to the original in terms of question interpretation.

Regarding CSI's internal consistency with its own factors, all correlations were significant and ranged between r= .42 and r= .68. When using the test-retest method, the correlation coefficient obtained between total scores was also significant, and r= .70. Validity evidence in relation to other variables was investigated by checking the correlation between the CSI and ^{Lipp's Stress Symptoms Inventory for Adults - LSSI (1984}). A total of 100 individuals who responded both instruments participated in this research. Results showed a low but significant correlation between the total scores of the two instruments (r= .14, p= .05). Considering these results, the authors concluded that the instrument has satisfactory psychometric qualities (^{Savóia et al., 1996}).

^{Dinis, Gouveia and Duarte (2011}) expanded the research on coping assessment and suggested a measure with four factors, namely: rational, emotional, avoiding and distant/disconnected. In addition, more recent studies have found correlations between coping strategies and positive personal characteristics, such as problem-solving skills and quality of life in patients with schizophrenia (^{Park & Sung, 2016}); positive emotions, resilience and mental health (^{Gloria & Steinhardt, 2016}); academic adaptation (^{Luca, Noronha, & Queluz, 2018}) and character strengths (^{Gustems-Carnicer & Calderón, 2016}). Character strengths will be used in this study to check the instruments validity evidence based on its relationship with related constructs (^{American Educational Research Association, American Psychological Association and National Council on Measurement in Education, 2014}).

Character strengths are defined as positive psychological traits that can push the individual forward, favoring a healthy development at the psychological, biological and social levels (Peterson & Seligman, 2004). In this sense, positive psychology proposes the existence of 24 strengths, namely: love, love of learning, appreciation of beauty, authenticity, self-regulation, kindness, bravery, citizenship, creativity, curiosity, hope, spirituality, impartiality, social intelligence, gratitude, humor, leadership, modesty, critical thinking, forgiveness, perseverance, prudence, judgment and vitality (Peterson & Seligman, 2004).

^{Harzer and Ruch (2015)} sought relationships between Character Strengths and coping in a study with 214 participants, 143 of whom were female, aged 21 to 64 years (M= 38.28 years; SD= 10.51). Intellectual, emotional and interpersonal strengths were positively correlated with positive coping strategies. The authors concluded that character strengths are related to coping behaviors and, therefore, function as a protective factor for stressful situations.

Another point that has been increasingly highlighted in more recent studies (^{Ben-Zur, 2019}; ^{Greenaway et al., 2015}) is the situational aspect of coping. In this sense, coping strategies should be assessed based on the context in which the stress episode occurs, the characteristics of the event itself, and the responses of the individuals involved, not by determining if strategies are good or bad, adaptive or maladaptive, which does happen in the CSI. The instrument proved to be a reliable measure (^{Savóia et al., 1996}), but its validity evidence in Brazil was obtained more than 20 years ago. The Standards for Educational and Psychological Testing (^{AERA, APA, & NMCE, 2014}) recommends that an instrument's validity evidence be constantly tested and updated to check if the measures are still reliable.

Given the above, it is important to check for new validity evidence for the CSI and assess whether its factor structure still holds. Therefore, the aim of this study was to analyze the psychometric properties of the Coping Strategies Inventory (Folkman & ^{Lazarus, 1984}), adapted to Portuguese by ^{Savóia et al. (1996}), through its validity evidence based on internal (factor) structure and relationship with other related variables (character strengths), as well as assessing the reliability of the instrument (^{AERA, APA, & NMCE, 2014}). As a guiding hypothesis, it was expected that the eight-factor structure found by ^{Savóia et al. (1996)} was confirmed; the overall and factor scores of the Coping Strategies Inventory were correlated with each other and with the Character Strength measures; and the CSI showed good reliability measures.

Method

Participants

A total of 927 college students participated in this study, with an average age of 26 years (SD= 7.66), ranging from 18 to 59 years, 61% were female, and the majority was single (69.3%). The data analyzed were collected from Psychology (47.9%), Engineering (36.4%) and Pedagogy (15.8%) students. Data collection was conducted at a private institution in the interior of the state of São Paulo (73%) and another in the state of Paraíba (27%). The demographic profile as show in Table 1.

Table 1: Participants Demographic Profile 

Results

Validity evidence based on internal structure

Table 2 shows the adjustment indices obtained after performing a confirmatory factor analysis of the Coping Strategies Inventory model proposed by ^{Savóia et al. (1996}). According to the results shown in Table 2, the proposed model did not present an adequate adjustment, since reference values were not within the expected range (^{Hu & Bentler, 1999}). Thus, an exploratory factor analysis (EFA) was performed with the Coping Strategies Inventory to check what would be the structure found in a current Brazilian sample.

Table 2: Adjustment indices of the Coping Strategy Inventory model proposed by ^{Savóia et al. (1996}) found in CFA. 

In EFA, it was first checked whether the data matrix was subject to factoring through the Kaiser- Meyer-Olkin (KMO) value, which was 0.90. Since this value was adequate (^{Damásio, 2012}), the next step was to perform factor retention. The extraction method was Robust Diagonally Weighted Least Squares (RDWLS), using the Promin rotation method. The number of factors was chosen according to the result of Parallel Analysis and the Hull Method (^{Ledesma, Ferrando, & Tosi, 2019}). Parallel analysis indicated a solution with 4 factors and the Hull Method indicated a solution with 2 factors. Then, EFAs were performed on both solutions to check which one was the best theoretical and statistical fit, considering the adjustment indices for each one. The exclusion criteria for each EFA were removing items that did not saturate at least .30 or that presented a similar saturation in more than one factor, until no items needed to be excluded.

In the first round of the two-factor structure, six items were removed, as five (3, 5, 7, 10, 36) did not saturate in any of the factors, and one (65) showed a similar saturation in both. In the second round, one item (62) was removed, because it showed a similar saturation in both factors. In the third round, no items needed to be removed. In the end, the two-factor structure showed 31.37% of total explained variance and all adjustment indices had adequate values (^{Hu & Bentler, 1999}), as shown in Table 3.

In the first round of EFA for the four-factor structure model, eleven items (3, 9, 29, 34, 35, 48, 54, 63, 64, 65 and 66) were removed because they did not saturate the minimum value in any factor, and four items were removed (10, 26, 47, 50) for having a similar saturation. In the second round, two items were removed because they had a similar saturation (51, 62). In the third round, two more items were removed, one for not saturating (25) and another for presenting a similar saturation in 2 factors (40). In the fourth round, item 43 was removed because it had a similar saturation in two factors. Finally, in the fifth round no items needed to be removed. In the end, the four-factor structure showed 40.64% of total explained variance and all adjustment indices had adequate values (^{Hu & Bentler, 1999}).

When comparing both models and considering the adjustment indices, total explained variance, and theoretical cohesion as proposed by ^{Folkman and Lazarus (1980}), the four-factor solution was chosen. Factors were renamed according to the theoretical proximity observed in groupings. The final structure found had 46 items. Factor 1 was interpreted as reflecting “Positive Reappraisal” (items 55, 56, 57, 58, 59, 60 and 61), Factor 2 as “Distancing and Acceptance” (items 4, 5, 6, 11, 12, 13, 14, 16, 17, 24, 32, 33, 37, 41, 44 and 53), Factor 3 as “Social Support” (items 7, 8, 22, 28, 31, 36, 42 and 45), and Factor 4 as “Confrontation and problem solving” (items 1, 2, 15, 18, 19, 20, 21, 23, 27, 30, 38, 39, 46, 49 and 52).

Table 3: Fit indices obtained in the EFAs performed in this study 

Table 4 shows the correlations between the total CSI score and its respective factors. All correlations were statistically significant. The highest correlations were those between the CSI Total Score and all its factors in general, which were strong. Factor 1 was moderately correlated with factors 2, 3 and 4, and Factor 3 was moderately correlated with factor 4. The lowest correlations were between Factor 2 and factors 3 and 4.

Table 4: Correlations between the Total Score of the Coping Strategies Inventory and the Scores of the Four Factors 

In this study, for Factor 1 (7 items), α= .79; for Factor 2 (16 items), α= .79; for Factor 3 (8 items), α= .67; and for Factor 4 (15 items), α= .86. The overall Cronbach's alpha (α= .89) was excellent. All these coefficients can be considered adequate reliability values (^{Marôco, 2014}).

Validity evidence based on relationships with other variables

Table 5 shows the correlations between the factors and the total CSI score with Character Strengths. It can be noticed that the Total CSI Score showed statistically significant, positive and weak correlations with Character Strengths, and did not show a significant coefficient only with Self-Regulation and Forgiveness. The CSI Positive Reappraisal Factor showed statistically significant and positive correlations with Appreciation of beauty, Kindness, Bravery, Citizenship, Curiosity, Spirituality, Social Intelligence and Judgment, and negative correlations with Self-regulation, all of which were weak. Distancing and Acceptance showed a statistically negative and weak correlation with most Character Strengths, except for Appreciation of beauty, Bravery, Creativity and Judgment. In the Social Support Factor, all Character Strengths were also positively and weakly correlated, except for a single Character Strength, Self-Regulation. Finally, Confrontation and Problem Solving was significantly and positively correlated with all Character Strengths, and most of these correlations were moderate.

Table 5: Correlations between the factors and the total CSI score with the character Strengths 

Discussion

The aim of this study was to analyze the psychometric properties of the CSI by checking its validity evidence, based on internal (factor) structure and relationship with other variables (character strengths), as well as assessing its reliability (AERA, APA, & NMCE). As a guiding hypothesis, it was expected that the eight-factor structure found by ^{Savóia et al. (1996}) was confirmed. However, when the model by Savóia et al. (1996) was tested, the adjustment indices found were not within the expected reference values. One hypothesis to explain this finding is the period that has passed between the testing of the model in 1996 and the current testing, indicating that some characteristic of the population may have changed. Regarding the sample, the study by Savóia et al. (1996) was conducted in the state of São Paulo, while this study was carried out both in São Paulo and in Paraíba. Therefore, the differences may be justified by the inclusion of a new state in data collection. Both studies were conducted with college students, however, Savóia et al. (1996) provide only the N of the sample, so it is not possible to make other comparisons or formulate hypotheses on potential reasons why the model was not confirmed. The authors did not provide data such as age, gender, whether the institution was public or private, socioeconomic characteristics, among other possibilities.

With regard to factor structure, since the model of ^{Savóia et al. (1996}) was not confirmed, a new exploratory factor analysis was performed to check what item factor structure would be found. After running the new exploratory factor analysis, the most adequate structure was a four-factor model: Positive Reappraisal, Distancing and Acceptance, Social Support, and Confrontation and Problem Solving. In the new solution, adjustment indices and reliability values were adequate, and factors were significantly correlated with each other and with the overall CSI score. This set of variables indicates that the new factor structure shows validity evidence based on internal structure (AERA, APA, & NMCE, 2014).

When comparing the factor structure obtained in this study with the structure of ^{Savóia et al. (1996}), it can be noticed that the factors Accepting Responsibility and Self-control were eliminated in the new exploratory factor analysis. For the Self-control factor, the only item that has remained was factor 14, which was relocated to the Distancing and Acceptance factor in the new version. Item 14, which says “I tried to keep my feelings to myself”, is theoretically justified in the factor that combines Distancing and Acceptance. One hypothesis to explain why Self-control items have been almost entirely eliminated is that individuals can use stress coping strategies without necessarily having self-control. Accepting Responsibility items, which involve recognition of guilt, are also requirements for the manifestation of stress coping strategies. Simply accepting that some situations that cannot be changed can be a strategy in itself, but responsibility, especially in the university context, requires further actions, which may justify the extinction of the factor (^{Greenaway et al. 2015}).

With regard to the other factors, Positive Reappraisal and Social Support were maintained, while Distancing, Acceptance, and Confrontation and Problem Solving, which were four factors, were combined and became two. Escape-Avoidance items were divided between Positive Reappraisal and Distancing and Acceptance in the current version. This is relevant because the Escape-Avoidance strategy may be related either to a process of distancing, when the individual does not want to deal with something, or to a process of reassessing the situation to understand if escape-avoidance is the best option to face the source of stress (^{Greenaway et al., 2015}).

After verifying the new CSI factor structure, it was assessed whether it showed validity evidence based on relationships with other variables (in this case, character strengths). Results showed that most character strengths were positively correlated with the total CSI score, indicating that coping strategies are related to the presence of character strengths. These data corroborate the study by ^{Gustems-Carnicer and Calderón (2016}) and Harzner and Ruch (2015), indicating that people with more character strengths also find it easier to develop coping strategies to deal with adverse situations.

With regard to factors, Factor 1 - Positive Reappraisal refers to the coping strategy aimed at controlling sadness-related emotions, as a form of reinterpretation, growth and personal change after a conflict situation (^{Lazarus & Folkman, 1984}b). It was the factor with the least significant correlations, all of which were weak. One explanation for this is that, while the factor refers to the ability to face sadness-related emotions, character strengths refer to positive personality characteristics, more associated with subjective well-being (Niemiec, 2014).

The CSI Distancing and Acceptance factor showed a statistically negative correlation with most character strengths. Behaviors related to Distancing and Acceptance were negative, as they refer to avoidance behaviors in stressful situations (^{Lazarus & Folkman, 1984}b), while character strengths are positive and contribute to help individuals and society to thrive (^{Seider, Jayawickreme, & Lerner, 2017}). Since they are negative, they can often become maladaptive, going in the opposite direction of the precepts of character strengths, because if adverse situations were faced, they could be circumvented (problem-focused coping) (^{Lazarus & Folkman, 1984b}). These indices suggest that, the more the strengths related to emotional qualities stand out, the better the coping strategies. McCrae and Costa (1986) reported that people with high levels of well-being use coping behaviors such as rational action, seeking help, drawing strength from adversity, and faith. Thus, coping theories are based on the idea that, in order to deal with problems, happy people trigger thoughts and behaviors that are adaptive and useful, while unhappy people, on average, face difficulties in a more destructive way (Diener, Suh, & Oishi, 1997; ^{Greenaway, 2015}).

The CSI Social Support factor was positively correlated with most character strengths. This result can be explained by the notion that people who have more character strengths are able to obtain more social support in their daily lives by becoming “easier” people to live with. Self-regulation had a coefficient equal to zero, which can be explained by the fact that it involves the individual's ability to behave effectively when facing adverse events, while social support involves other people, which does not necessarily occur when the person uses self-regulation in a given situation. The previous example fits here, a college student may decide to wake up early every day to study and do well in a test (behavioral self-regulation and problem-focused coping), but if they do it individually, without the help of others, they will have no social support, which justifies the lack of correlation with this character strength in this specific case.

The Confrontation and Problem-Solving factor showed a statistically significant correlation with all character strengths, being the factor with the highest number of moderate correlations, which makes theoretical sense, as it reflects the act of confronting a stressful situation. Therefore, this correlation indicates that the more developed the character strengths, the more individuals can manage adverse situations. These data confirm the studies by Harzner and Ruch (2015) and indicate that people who have stronger character strengths also find it easier to develop problem-solving and confrontation strategies. Finally, given the above and considering the results obtained, it can be postulated that the instrument also presents validity evidence based on relationships with other variables (AERA et al., 2014), since most character strengths were correlated with CSI measures, confirming the initial hypothesis of this study.

The character strengths that did not show any correlation were Self-Regulation and Forgiveness. Emotional self-regulation is the ability to moderate attention and behaviors arising from different circumstances and events (Peterson & Seligman, 2014). In this sense, it seems that participants may have coping strategies in their repertoire, but they have difficulty in self-regulating their behavior. For example, a student may use coping strategies such as praying, crossing fingers, thinking positive, among others, to try to do well in an exam, but not be able to get organized to study (behavioral self-regulation). To some extent, this information corroborates the exclusion of Self-control, as these factors have a similar content. Thus, in addition to its items not being retained in the source factor (^{Savóia et al., 1996}), the correlation between self-regulation and the total CSI score was null.

With regard to Forgiveness, this result also makes theoretical sense, as a person may have difficulty in forgiving others, but at the same time have numerous coping strategies in their daily repertoire. In addition, considering the study sample, a college student who cannot forgive their classmate can probably develop coping strategies to do well in the academic context, such as waking up early to study, organizing group work, and thinking positive (^{Gustems-Carnicer & Calderón, 2016}).

Final Considerations

Considering the objective of this article was to investigate the validity evidence of the Coping Strategies Inventory, results point to important evidence indicating that the data are relevant to understand the college students’ sample. Validity evidence based on internal structure and relationship with other variables was found. However, the structure did not corroborate the original version of the instrument, raising a new factor structure suggestion, which is one of the main contributions of this study. Furthermore, this study contributed to updating the instrument's format, as the last study aimed at checking its psychometric properties was conducted more than twenty years ago (^{Savóia et al., 1996}).

However, even considering these positive results, the data must be taken with reservations, since data collection was conducted only with college students, which is a limitation. Therefore, a suggestion for future studies would be to expand the investigation scope beyond the academic context and with greater age variability. It would also be important to check if the factor structure remains the same for other samples. Another possibility would be to collect data in different ways, not only through self-administered questionnaires, such as conducting interviews to include participants with less or no education who would not be able to respond without assistance.

Finally, it would also be important to analyze the relationship between CSI scores and other external variables, such as those already established in the literature: stress, depression, anxiety, among others. Given the above, results indicate that this study has contributed to update the validity evidence of CSI and that further research is needed to expand this evidence.