Since its introduction in 1990 (Bakker, 2017), work engagement has been an inspiration for researchers and practitioners around the world (e.g., Shimazu et al., 2008; Seppälä et al., 2009; Balducci et al., 2010; Nerstad et al., 2010; Fong and Ng, 2012; Littman-Ovadia and Balducci, 2013; Panthee et al., 2014; Zecca et al., 2015; Lovakov et al., 2017). Moreover, it has been regarded as a “societal challenge” (Schaufeli and De Witte, 2017, p. 58) with far reaching effects on the economy. It is widely accepted that, in order to be competitive, contemporary organizations need engaged employees (Bakker, 2017).

As suggested by Schaufeli (2017), based on analyzing the 6^th European Working Conditions Survey data from 35 countries, work engagement should be studied not only at the individual level, but also at the national level. Comparing countries with different levels of engagement, Schaufeli demonstrated that work engagement was positively related with national economic activity and productivity. On the European engagement map, the Netherlands, the originating country of the mostly used and cited theoretical model and instrument for assessing work-engagement, the Utrecht Work Engagement Scale (UWES, Schaufeli and Bakker, 2003), was the most engaged country. On the other side, Serbia, the country where the current research takes place, was the least engaged among the 35 surveyed European countries (Eurofound, 2016).

Since the 1990s, Serbia has been encountering a deep, long-lasting socioeconomic crisis, strongly influenced by the wars in the former Yugoslavia, UN sanctions and political instability. On the economic scene, the country went through several models of privatization, including a high failure rate of approximately 25% of unsuccessful privatizations (Vujačić and Petrović-Vujačić, 2011). Similar to other world countries, Serbia was exposed to the deep global financial crisis that had a negative impact on employees and their health and well-being across the world (Giorgi et al., 2015; Mucci et al., 2016; Lopez-Valcarcel and Barber, 2017). Continuously struggling with the hostile conditions, workplace in Serbia has been fluctuating with the numbers of new organizations, new business owners, either as a part of privatization or green-field investment, different work contracts, new technologies and work processes. Nevertheless, contemporary Serbian economy is still characterized by low economic activity (Petrović et al., 2017). GDP per capita is among the bottom 6% of the countries covered by the 2015 European Working Conditions Survey (Eurofound, 2016). The unemployment rate is 18.2% among population aged 15–64 (Statistical Office of the Republic of Serbia, 2016). Bearing all this in mind, in this paper, we deal with work engagement concept in the specific Serbian socio-economic context.

Work engagement is defined as a positive, fulfilling, affective-motivational state of work-related well-being (Bakker et al., 2008). It is characterized by vigor – higher levels of energy, mental resilience and investment of effort; dedication – involvement in work and the sense of meaningfulness and enthusiasm, and absorption – full concentration and engrossment in work (Schaufeli et al., 2002, 2006; Bakker, 2017). In a nutshell, work engagement is about giving “hands, head, and heart” at work (Ashforth and Humphrey, 1995, p. 110).

The concept of work engagement is deeply embedded in job demands–resources theory (JD–R), which sees engagement as a function of an interplay between job demands and resources (Bakker and Demerouti, 2007; Bakker, 2017). Work engagement is, thus, a mediator between job demands/resources and job performance. Job resources such as career opportunities, organizational support, job security, and positive work climate lead to high work engagement and, consequently, to outstanding performance. On the other hand, job demands such as high work pressure, poor physical conditions, negative work climate and challenging interpersonal relations are associated with negative psychological consequences. If a person does not pull together the personal and organizational resources in order to recover from pressures at work, job demands can lead to burnout and exhaustion.

The Utrecht Work Engagement Scale (UWES) developed by Arnold Bakker and Wilmar Schaufeli is the most frequently used measure of work engagement (Bakker and Demerouti, 2016). Systematic literature review shows the UWES was almost exclusively applied as a valid basis for developing work engagement interventions (Knight et al., 2017). The UWES is a self-report scale, with 17-item and 9-item versions widely used in independent national research studies all over the world. There is also an emerging three-item version that was applied in the 6^th European Working Conditions Survey (Schaufeli, 2017). Both long and short versions of the UWES cover all three theoretically postulated dimensions: vigor, dedication and absorption. In the 9-item version of the UWES each dimension is covered by three items, while in the 17-item version vigor is covered with six items, dedication with five items and absorption with six items (Schaufeli et al., 2002). Since two items of the original 17-item version were considered as problematic (item 6 from the vigor subscale, and item 6 from the absorption subscale), the 15-item version of the UWES has also been applied in research (Schaufeli and Bakker, 2003). The newest three-item version of the UWES covers all three dimensions of work engagement with one item each (Schaufeli, 2017): for vigor ‘At my work I feel full of energy’ (VI1, modified); for dedication ‘I am enthusiastic about my job’ (DE2), and for absorption ‘Time flies when I am working’ (AB1). There are 30 language versions of the UWES, available at the Schaufeli’s web site.

Previous national validation studies gave somewhat conflicting results about the structure of the UWES and its demographic correlates, while findings about well-being correlates have been quite consistent. Besides the initial Dutch version (Schaufeli and Bakker, 2003), confirmatory factor analysis supported theoretical conceptualization of the three-factor structure (vigor, dedication and absorption) for both the UWES-17 and the UWES-9 Finnish (Seppälä et al., 2009) and Norwegian (Nerstad et al., 2010) language versions. The UWES-9 three-factor structure was confirmed in Chinese (Fong and Ng, 2012); French (Zecca et al., 2015); Hebrew (Littman-Ovadia and Balducci, 2013); Italian (Balducci et al., 2010); Nepali (Panthee et al., 2014); Russian (Lovakov et al., 2017), and Swedish (Hallberg and Schaufeli, 2006) language versions. The one factor UWES-17 was found in the Japanese validation study (Shimazu et al., 2008). Moreover, it is interesting to note that based on the data from a large single company sample from South Africa, applying six UWES-17 language versions, i.e., Afrikaans, English, Nguni, Tshivenda (Venda), Sotho and Xitsonga (Tsonga), based on item response modeling, the one-factor solution has also been suggested (Goliath-Yarde and Roodt, 2011; de Bruin et al., 2013). The one factor UWES-9 solution was found in the Brazilian (Ferreira et al., 2016); Chinese (Fong and Ng, 2012); Japanese (Shimazu et al., 2008); Russian (Lovakov et al., 2017), and Swedish (Hallberg and Schaufeli, 2006) UWES versions. By analyzing the data from 10 different countries, Schaufeli et al. (2006) concluded that work engagement was weakly positively related with age, with correlations about 0.15 or less. Furthermore, the differences between men and women in relation to work engagement were ambiguous, with the lack of differences in some countries (Australia, Canada, and France), and low significant differences with low effect size in other countries (e.g., women from Netherlands, Spain, and South Africa showed higher engagement, whereas men from Belgium, Germany, Finland, and Norway expressed higher engagement than women). Managers scored higher on engagement than general staff (Andreassen et al., 2012; Littman-Ovadia and Balducci, 2013; Panthee et al., 2014).

Of all well-being measures, work engagement is usually linked with burnout. There are two different approaches regarding this relationship (Schaufeli and Bakker, 2003). The first one postulates that burnout is the opposite pole of work engagement (Maslach and Leiter, 1997). The second proposes that these constructs should be regarded as different (Schaufeli and Bakker, 2003). Schaufeli and Bakker (2003) support their argument by the thesis that the correlation between these concepts is not full, and that evaluating both concepts with the same instrument is not empirically justified. Some later tendencies postulate the thesis that burnout and engagement constitute ‘dual unity,’ which is at the same time ‘real and redundant’ (Schaufeli and De Witte, 2017, p. 58). Apart from burnout, work engagement factor structure is usually studied in relation with job satisfaction (e.g., Shimazu et al., 2008; Panthee et al., 2014; Lovakov et al., 2017).

The aim the current study was to analyze the psychometric properties of the UWES-S, the Serbian adaptation of the Utrecht Work Engagement Scale (UWES, Schaufeli and Bakker, 2003). In particular, the aims were to: (1) Investigate factorial structure, both for the 17-item and 9-item versions of the UWES-S; (2) Verify theoretically based structure of the long and short versions of the UWES-S, i.e., to compare the theoretically based three-dimension model of vigor, dedication, and absorption with the one-dimension solution supported by some validation studies (e.g., Shimazu et al., 2008); (3) Investigate Cronbach’s Alpha reliability of the different versions of the UWES and corresponding subscales; (4) Investigate the construct validity through the relationship between work engagement and selected well-being indicators (burnout, organizational support, job satisfaction, job insecurity), and (5) Explore work engagement across demographic variables. By doing so, we wanted to deepen our understanding of work engagement by offering an insight from the Serbian cultural and economic setting, specifically different from the Western-European UWES originating setting, and the settings where mainstream work engagement research and validation studies took place.

The overall aim of the present study was to provide evidence of the psychometric properties of the Serbian versions of the UWES, the long and short one, by exploring their factor structure, internal consistency and criterion validity.

A detailed analysis confirmed that both the UWES-17 and UWES-9 were applicable in the Serbian cultural context. Based on the UWES-17 findings, our data have confirmed both the three-factor and one-factor solutions, giving a slight advantage to the three-factor solution. As for the UWES-9, based on the PCFA and CFA findings, we obtained one-factor as a preferred solution. Based on a systematic review of research papers dealing with the UWES factorial validity within the CFA approach, the one-factor solution of the UWES-9 was also suggested by Kulikowski (2017). Taking into the account all other analyses presented in this paper, we would opt for applying both one and three-factor solutions for the UWES-9 Serbian version.

The PCFA of the 17-item version of the UWES-S showed a three-factor structure with the content somewhat different from the original model. The first component was majorly loaded by dedication items (including absorption items), the second component was loaded by vigor and absorption items, while the third component was clearly loaded only by vigor items. The most “problematic” were the absorption items that actually spread across two dimensions. The first component pointed to involvement in one’s job, the second could indicate general activity, while the third component indicated stamina. The item that was particularly problematic was the DE2 item (“I am enthusiastic about my job”) that loaded almost equally on both the first and the second component. The PCFA of the 9-item version showed the existence of only one component, with loadings above 0.590.

Overall, CFA revealed better fit indices for the UWES-17 than the UWES-9. The long version of the UWES showed the best fit when three factors were taken into account. The fit indices were somewhat better for the structure that was obtained by PCFA, but the indices for the original three factor model were also acceptable. The UWES-9 showed a somewhat better fit when the one-factor solution was tested. Nevertheless, taking into consideration the fit indices of all tested models, we can conclude that we only gained an acceptable, but not a preferred fit. Even though these indices are acceptable when comparing with other international samples (e.g., Schaufeli and Bakker, 2003; Schaufeli et al., 2006), this finding could be also interpreted as an “invitation” for making some culturally more sensitive forms of the UWES. Bearing in mind qualitative data from the focus group discussion about understanding the UWES items, the presented data suggest it is reasonable to question whether the employees from Serbia conceptualize work engagement similarly to Dutch employees. In a broader perspective, obtained results point to the need for more extensive and precise multicultural comparisons, which is in line with Balducci et al.’s (2010) conclusions about the need for more stringent cross-cultural research.

Contrary to the ambiguities brought by PCFA and CFA, the reliabilities, inter-correlations and the pattern of correlations of the UWES (short and long) with the corresponding well-being variables clearly met all theoretical expectations. The reliabilities of the UWES-17 and UWES-9 total scores were higher than 0.90. All the originally postulated subscales on both forms, except for the UWES-9 absorption, showed reliability above 0.78. The inter-correlation between subscales of both forms showed that the corresponding sub-scales shared more than 79% of variance. In addition, the correlation of both UWES forms with burnout, job satisfaction, organizational support and job insecurity showed an almost equal pattern of correlations when analyzing total scores and the scores on particular subscales. Finally, when taking into account the incremental validity of work engagement above job burnout, assessed by both the UWES-17 and the UWES-9, in both cases, when UWES was included, it gave better prediction of job satisfaction with almost equal values of all parameters. These positive and theoretically expected findings indicate that both the UWES-17 and the UWES-9 can be used with equal success in research of work engagement in Serbia.

Observed relationships between work engagement (assessed both by UWES-17 and UWES-9) and demographic variables add to the muddled picture of the previously published correlations that are clearly inconsistent (e.g., Schaufeli et al., 2006; Fong and Ng, 2012; Littman-Ovadia and Balducci, 2013; Lovakov et al., 2017). Generally, when comparing the employees’ levels of engagement across gender, education, hierarchical position, as well as the ownership of organization, it should be noted that observed statistically significant differences have fallen under the shadow of the low effect size. As underlined by Schaufeli et al. (2006), these differences lack practical significance.

The obtained results are more interesting and valuable because they come from the Serbian socio-economic context, in which, for the past quarter of the century, employees have been confronted with privatizations, restructuring, downsizing, closing down and the collapse of a great number of companies (Petrovic and Cizmic, 2010). For many it meant either losing their job or early retirement. Losing a job subsequently led to long-term unemployment and exponentially declining chances of finding a job. As for the employees who stayed at work as the ‘survivors’ of organization and society-wide negative processes, it is questionable whether staying at work sustained their work engagement or eroded it. As noted in our previous research, the mechanisms which helped the workers to cope with the crisis could be a serious threat for readapting to working under the ‘normal’ conditions (Čizmić et al., 1995). The third work engagement component obtained by PCFA, loaded by three vigor items, named ‘stamina,’ could be related to the specificities of the Serbian social and economic context. Further, different factor solutions yielded based on different UWES language versions could be an impulse for exploring cultural specificities of work engagement and searching for culturally invariant factors of the UWES.

Some potential limitations may exist in the presented research. Although we had a large sample from different companies, economy sectors and occupational groups that were chosen to represent Serbian workforce well, the sample was not composed as statistically representative. It can pose the limitation for comparing the presented findings with the findings of other validation studies, more so because the majority of these studies were also not carried out on statistically representative samples. The cross-sectional nature of the study can also be regarded as a limitation for generalizing and fully utilizing the findings that at present do not give grounds for inferring about the directions of the presented relationships. As suggested by other researchers, future longitudinal studies should uncover causal relationships of work engagement with other well-being correlates (e.g., Schaufeli et al., 2006; Shimazu et al., 2008; Fong and Ng, 2012). Moreover, as noted by Littman-Ovadia and Balducci (2013), the restriction of range in this kind of research design could be expected for work engagement and especially in the Serbian work context. The restriction of range should also be taken into account in those countries with high rates of emigrating workforce. Keeping in mind that work engagement is one of the building blocks of the JD–R model, to develop a full understanding and validation of the model, engagement should be more fully explored in the relationship with other features of the model. Finally, as noted by some researchers (e.g., Fong and Ng, 2012), self-report nature of applied measures is prone to common method variance. Thus, future studies could include some more objective measures of work engagement.

The current study highlights the validity of the UWES in the specific social and economic context of Serbia. The presented results have shown that both the UWES-17 and the UWES-9 Serbian versions have satisfactory psychometric properties with high reliability, factorial structure in line with theoretical model and good predictive validity, thus confirming that these UWES versions are applicable in the Serbian context, both for use in research and for practical purposes. Nevertheless, there is a need for exploring how employees from Serbia as well as other countries perceive work engagement. As suggested by PCA, it is reasonable to question whether the employees from Serbia conceptualize work engagement in the same way as the workers from other cultures and economies. Future cross-cultural research should further investigate the cultural invariance of the UWES factorial structure under more controlled conditions. The presented research could be regarded as a contribution to a positive approach to evolving international economic cooperation. Shedding light on work engagement contributes to a positive psychology approach that is essential for successful work transformation and organizational transitions in globalizing economy.

All three authors, IP, MV, and SČ contributed equally to the research design and writing of this paper.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.