Measures of positive psychological constructs are considered important in understanding the personal resources that help employees cope constructively with work–life stressors in today’s fast-paced and more turbulent work environment (Görgens-Ekermans & Herbert, 2013; Metzer, De Bruin & Adams, 2014; Sims, Hogan & Carstensen, 2015). The field of positive psychology focuses on uncovering psychosocial behavioural strengths (rather than deficits) that contribute to optimal positive human functioning (Snyder, Lopez & Pedrotti, 2011). Positive coping strengths are important personal resources in helping employees deal constructively with the complex interaction between the individual and the work environment (Demerouti & Bakker, 2011). Constructs relating to positive emotion (e.g. emotional intelligence, positive affect, happiness), cognition (e.g. hope, optimism, problem-solving, self-acceptance, self-esteem), intrinsic motivation (e.g. locus of control, resilience, self-efficacy, flourishing) and interpersonal interactions (e.g. extroversion, agreeableness, social support) are inherent to positive coping behaviour and seen to facilitate better adjustment to stressful life events (Hammer, 1988; Marx, 2016; Proyer, Gander, Wellenzohn & Ruch, 2014). Measuring individuals’ positive coping strengths in a valid and reliable manner has therefore become an important endeavour for scholars and practitioners as studies have shown important links between positive coping constructs and positive work and organisational outcomes such as work engagement, job satisfaction, productivity, well-being and low turnover intention (Mayer, 2014; Rothmann, 2014; Van der Heijde, 2014). The present research contributes to the emerging South African–based research on positive psychology in its aim to examine the construct validity of the positive coping behavioural inventory (PCBI) that was developed by Marx (2016) for the South African work context.

The present study contributes to the emerging positive psychology research literature by examining the reliability and validity of the factor structure identified by Marx (2016) with the view to further refine the PCBI scale and increase its usefulness and application value in the South African workplace. The PCBI is seen to fill an important gap in the research literature by offering a multidimensional measure that assesses a wide array of positive psychological constructs that have been indicated by the research literature as important positive strengths in coping with stress and improving employee well-being (Marx, 2016). Further examination of the mutidimensional psychometric properties of the scale is therefore warranted and important.

Marx (2016) approaches positive coping behaviour from a multidimensional and psychosocial perspective. As shown in Table 1, the psychosocial dimensions of positive coping behaviour (Marx, 2016) relate to positive behavioural capacities embedded in individuals’: (1) inventive coping behaviour (positive problem-solving anchored in individuals’ self-esteem, internal locus of control, optimism, humour and positive reframing, emotional, motivational and interpersonal [social] capacities), (2) engaging coping behaviour (generating and maintaining positive emotions, feeling self-efficacious in alleviating stress through emotional granularity and feelings of happiness and engagement), (3) intentional coping behaviour (proactive coping through intrinsic motivational and self-efficacious goal-directed behaviour, conscientiousness, adaptability, resilience and flourishing) and (4) influential coping behaviour (positive social behaviour through extroverted behaviour, positive interactions with others, agreeableness and building social support). This multidimensional approach towards positive coping is in agreement with the argument that coping resources consist of a complex and dynamic set of cognitive, emotional, motivational and social (interpersonal) behavioural responses that affect the outcome of potentially stressful events (Hammer, 1988; Papastavrou et al., 2011).

Scholars engaged in positive psychology research generally argue in favour of multidimensional measures of well-being to understand positive coping behaviour (Luthans et al., 2007; Rothmann, 2014). Advantages of measures with multidimensional constructs are that they can help to capture the complexity of positive coping behaviour and well-being and help to identify the strengths and potential areas for growth that inform intervention design (Rothmann, 2014).

Flourishing theory (Keyes & Annas, 2009; Ryff & Keyes, 1995; Youssef & Luthans, 2012) postulates that optimal human flourishing is characterised by growth and generativity that stem from mental health (thinking well), emotional well-being (feeling well) and psychological and social well-being (functioning well). Seen through the theoretical lens of flourishing theory (Keyes & Annas, 2009; Ryff & Keyes, 1995), it is argued that positive coping behaviour represents an array of responses flowing from a positive state of mental health and emotional, psychological and social well-being. Each state of optimal flourishing (i.e. mental, emotional, psychological, social) is supported by an array of self-regulating personal resources that affect the outcome of potential stressful events that the individual has to confront in the person–environment interaction process (Rothmann, 2014). Marx’s (2016) theoretical framework and empirical measure (PCBI) organise a wide array of positive personal coping and flourishing resources into four dimensions of positive coping behaviour (i.e. inventive coping behaviour, engaging coping behaviour, influential coping behaviour and social coping behaviour). The common theme underlying these four dimensions is the link with individual mental health and emotional, psychological and social well-being as espoused in flourishing theory.

Research indicates that mental health is supported by cognitive orientations and traits such as wisdom, self-esteem, humour, internal locus of control, openness to experience and positive reframing. These positive coping constructs have been associated with well-being and creative (inventive) problem-solving behaviour (Ardelt & Edwards, 2015; Helmer, Krämer & Mikolajczyk, 2012; Kleiman & Riskind, 2012; Lambert, Graham, Fincham & Stillman, 2009; McGraw, Warren & Kan, 2015; Ong, Phillips & Chai, 2013). Positive emotions, emotional granularity and happiness are positively linked to adaptive coping behaviour (Serena, 2013), engagement (Sulea, van Beek, Sarbescu, Virga & Schaufeli, 2015), mental health (Snyder et al., 2011), and well-being (De Zutter et al., 2010). Psychological well-being constructs such as self-efficacy, resilience, flourishing, intentions for positive health, proactive coping, conscientiousness and adaptability are positively associated with intrinsic motivational and intentional goal-directed behaviour and the capacity to adapt to stress and adversity (Asamani, Cobbold & Dai-Kosi, 2015; Kim, Newton, Sachs, Glutting & Glanz, 2012; Marx, 2017; Stratta et al., 2015; Windle, 2011; Zhao & Seibert, 2006). Positive traits such as extroversion and agreeableness are positively linked to positive coping and the capacity to build social support (Fornés-Vives, Garcia-Banda, Frias-Navarro & Rosales-Viladrich, 2015; Lucas & Fujita, 2000; Sulea et al., 2015). The value of Marx’s (2016) initial research is the integration of a wide array of positive psychology constructs that have emerged as important flourishing and well-being constructs into a higher-order four-dimensional theoretical framework and empirical measure (PCBI) of positive coping behaviour. Although the initial EFA conducted by Marx (2016) provided evidence of the four-factor structure of the postulated four-dimensional theoretical framework, further validation of the internal multidimensional structure of the PCBI is still lacking. The present study addresses this gap in research by examining the construct validity and internal consistency reliability of the PCBI.

The research followed a cross-sectional quantitative research approach. Non-probability convenience sampling was utilised to collect primary data.

The initial four-factor best fit CFA model was examined in terms of descriptive statistics in order to identify items and factors with low internal consistency reliabilities. The descriptive statistics (means and standard deviations, internal consistency reliability coefficients, AVEs) and bivariate correlations are reported in Table 2. In terms of the means, low scores imply positive self-evaluation about the capacity to demonstrate the behaviour while high scores imply negative self-evaluation about the capacity to demonstrate the behaviour (Marx, 2016). Overall, the mean scores reflected positive self-evaluations of the capacity to demonstrate positive coping behaviour. The internal consistency reliability coefficients (CRs) were all well above the threshold value of 0.70 (acceptable) with a range between 0.82 and 0.95. The AVEs were below the threshold value of 0.50 (range: 0.37 to 0.44), which indicates the presence of measurement error.

The bivariate correlations among the four subfactors were all positive and significant (r ≥ 0.63 to r ≤ 0.78; p = 0.0001; large practical effect). All four factors correlated positively and significantly with the overall PCBI construct (r ≥ 0.87 to r ≤ 0.96; p = 0.0001; large practical effect).

The first step was to test for potential common method bias because of the cross-sectional nature of the research design and the subjective self-report approach of the PCBI. The Harman’s one-factor solution for the PCBI showed that loading the original 41 items onto one overall construct accounted for only 17.69% of the covariance among the scale variables. The one-factor CFA solution (all items of the PCBI loading onto one-factor) had poor data fit with CFI and NNFI values below 0.80 (see Table 3). These results suggested that common method bias did not pose a serious threat to the findings.

The CFA one-factor model was compared with a four-factor CFA model based on the original 41-item PCBI. As reported in Table 3, the original four-factor solution had less than acceptable data fit with the CFI and NNFI values being lower than the threshold value of 0.90 (CFI = 0.88 and NNFI = 0.87). However, the four-factor solution (all items loading onto the respective subfactor, and the four factors loading onto the overall PCBI construct) had better model fit than the one-factor CFA solution with a lower AIC value and improvement in the RMSEA (0.06) and SRMR (0.05) values which indicated reasonable fit. These results provided initial evidence of internal construct (convergent and discriminant) validity of the four-factor PCBI.

Model modification of the four-factor model was then done in order to improve the CFI and NNFI fit indices. This step involved removing two problematic items (path coefficients below 0.50) from the inventive coping behaviour subscale. This step reduced the original 41 items to 39 items. Table 3 shows that the modified four-factor solution had only a marginal model improvement with lower RMSEA (0.05) and SRMR (0.04) and AIC (1734.5917) values, but with the CFI (0.89) and NNFI (0.88) being still lower than the threshold value of >0.90.

Inspection of the modified four-factor solution model showed that although the influential coping subscale had a strong loading (0.87) onto the overall PCBI construct, its loading was the lowest in comparison with the other three subfactors (>0.94). The influential coping subscale also had the lowest correlations with the other three factors and overall PCBI scale (see Table 2). It was therefore decided to exclude the 5-item influential coping subscale in order to improve the model fit of the PCBI. This step reduced the modified 39-item PCBI to only 34 items.

The third CFA model (34 items) involved calculating a three-factor CFA solution (items of the three PCBI factors – inventive coping behaviour, engaging coping behaviour and intentional coping behaviour – loading onto the three factors, and the three factors loading onto the overall PCBI construct – see Table 4). As shown in Table 3, the three-factor CFA solution showed model improvement and good model fit (lower AIC = 1429.5063; CFI = 0.92 and NNFI = 0.90; RMSEA = 0.05 and SRMR = 0.04). Table 4 shows that all path coefficients were significant (t > 2.56; p ≤ 0.01) and above the threshold value of 0.50. These results provided evidence of the structural (convergent) validity of the three-factor CFA solution of the PCBI. Analysis of the internal reliability coefficients of the three-factor model revealed high reliability coefficients: inventive coping: α = 0.83/CR = 0.88; engaging coping behaviour: α = 0.87/CR = 0.91; intentional coping behaviour: α = 0.93/CR = 0.95; and overall PCBI: α = 0.95/CR = 0.96.

Closer inspection of Table 4 revealed that the inventive coping behaviour sub-dimension retained eight items relating to positive psychology constructs such as wisdom, self-esteem, optimism, humour, locus of control, openness to experience and positive reframing. The eight items of the engaging coping behaviour factor related to positive coping constructs such as positive affect, emotional granularity, happiness and self-efficacy. The 18 items of the intentional coping behaviour sub-dimension related to positive coping constructs such as self-efficacy, resilience, flourishing, intention for positive health, proactive behaviour, conscientiousness and adaptability.

Because of the cross-sectional design of the research, the results cannot be generalised to other industry settings. It is therefore imperative that the study be replicated in relation to other measures of positive psychology construct measures across a wider, more representative South African sample in various industry settings. The low AVE values indicated the presence of measurement error, which was not further investigated in the study. Consideration of confounding factors that could influence self-report measures such as the PCBI and therefore contribute to measurement error should be incorporated in future research endeavours.

This study contributed to the emerging body of knowledge on the assessment of positive psychology constructs that contribute to employees’ well-being and flourishing in the South African workplace. The results provide preliminary evidence of the usefulness of the PCBI as a valid and reliable multidimensional measure that integrates a wide array of positive psychology attributes in a single measure. The convergent validity of the PCBI as a measure of three higher-order dimensions of positive coping behaviour (inventive, engaging and intentional coping behaviours) was demonstrated in this study. Researchers may confidently use the three-factor solution of the PCBI to measure employees’ self-evaluations of their capacity to demonstrate positive coping behaviour in the workplace.