In the long tradition of stress research, the terms stress and strain are often used interchangeably. A strict way to differentiate between them was introduced in the norm series of the ISO 10075 (ISO, 2000a,b, 2015). As we can see norms as the current status of scientific knowledge we therefore refer to these definitions in this text. The norm ISO 10075-1 (ISO, 2000a, p. 3; s.a. Demerouti et al., 2002; Zoni and Lucchini, 2012), defines mental stress and strain as follows: mental stress is the “…total of all assessable influences impinging upon a human being from external sources and affecting that person mentally” and mental strain is defined as the “…immediate effect of mental stress within the individual depending on their current condition.” Thus, the norm summarizes the objective stressors under the term “stress” and the outcome of these stressors (e.g., the individual evaluation of these stressors) as “strain.” The consequences of strain can be further differentiated into short-term and long-term effects of strain. Short-term effects compass mental fatigue, monotony, satiation and stress sensations (Demerouti et al., 2002). Long-term effects of strain result from repeated exposure to strain. One important chronic reaction to prolonged impairing strain would be burnout which is now mentioned especially in the upcoming norm of the ISO 10075-1 (ISO, 2015).

The definition of stress is similar to the definition of “job demands” (Bakker et al., 2005), as job demands are described as physical, mental, social, or organizational aspects of the job that require sustained effort/skills and are associated with physiological and psychological costs (as in the concept of Meijman and Mulder, 1998). Bakker et al. (2005) emphasize that job demands are not necessarily negative and can influence positive outcomes (e.g., engagement) as well.

Stress is a neutral term and does not bear a negative (or positive) connotation. In the ISO 10075-1 (ISO, 2000a), stress is seen as a total value that results by summarizing all influences at the workplace – whether they have negative effects or not. However, it is possible to distinguish different stressors that might have different short-term or long-term outcomes. Rohmert (1984) distinguishes quantifiable and non-quantifiable stressors. In his view, quantifiable stressors compasses noise, temperature or air pressure. Non-quantifiable stressors usually results from working tasks and includes time pressure, monotony or responsibility.

One classification of stressors is stated in the ISO 10075-1 (ISO, 2000a, p. 4), which categorizes stress in following categories: task requirements, physical conditions, social and organizational factors, and societal factors. The ISO concept therefore focuses to a system of stressors which can lead to a variety of possible outcomes, negative but also positive ones. The main strength of this model lies in the strict distinction between stressors and the effects. As positive outcomes (e.g., practice effects or competence development) are also possible it is not meaningful to classify stressors as “good” or “bad,” the consequences of stress are “depending on an individual’s personal resources and his/her perception of the situation” (Demerouti et al., 2002, p. 425).

A framework that compasses different stressors which are seen as potential risk factors especially for burnout can be seen in the concept of the areas of worklife (Maslach and Leiter, 2008). In this concept, six domains of the work environment have been identified that can serve as “organizational risk factors” (Maslach and Leiter, 2008, p. 500) for negative work-related outcomes, such as burnout: workload, control, reward, community, fairness, and values.

The area of workload is described as experiencing qualitative and quantitative work overload that depletes the person’s capacity to meet the demands of the job. Control means having sufficient latitude at work and having possibilities to make important decisions. Reward refers to financial and non-financial reward (e.g., recognition from colleagues, supervisors and clients) for the employees’ work contributions. Community is the overall quality of social interactions of work. The area of fairness describes the extent to which decisions at work are perceived as fair. Finally, values (or value-congruence) describes the match of the employees’ and organizations’ job goals and expectations (a full description can be found, e.g., in Maslach and Leiter, 2008).

In all six areas, discrepancies between person and work environment can occur which are described as mismatch (Maslach and Leiter, 2008). In other words, these six areas are able to reveal critical work conditions. On the other hand, they serve as prevention factors for negative work-related outcomes if they are evaluated positively (Laschinger et al., 2015; Boamah and Laschinger, 2016). Especially the area of values is found to have an essential effect on work-related outcomes such as burnout (emotional exhaustion, cynicism, and personal accomplishment) and turnover intention (Leiter and Maslach, 2005, 2009; Leiter et al., 2009). A conflict between individual and organizational values leads to less engagement with the tasks, as work is perceived as personally irrelevant (Leiter and Maslach, 1999). This reduced involvement depletes the employees’ energy and contributes to exhaustion and cynicism.

The areas of workload and reward are important predictors for emotional exhaustion and cynicism, respectively (Leiter and Maslach, 2009). A lack of (financial or non-financial) reward leads to feelings of inefficacy and meaninglessness and this in turn contributes to cynicism (Cartwright and Holmes, 2006). High workload prevents employees from adequate recovery which is a critical factor for fatigue and exhaustion (Sonnentag et al., 2010). Especially the relationship between workload and emotional exhaustion is rather persistent and stable over time, which means that high workload at one point is able to predict emotional exhaustion even after a long period of time (Michielsen et al., 2004).

The extent of short-term and long-term outcomes of stress depends on moderating variables, such as personal or workplace resources. In the ISO 10075-1 (ISO, 2000a), the current condition of the individual is highlighted, which is dependent on the individual’s workplace resources. Next to individual resources, workplace resources (or “job resources”) have received much attention in research about outcomes of stress (Karasek, 1979; Bakker et al., 2005). Indeed, targeting workplace resources is a sustainable way to prevent negative outcomes of stress as workplace resources seem to be the precursor of individual resources such as employees’ self-efficacy, self-esteem, and optimism (Xanthopoulou et al., 2007).

Resources in the working context refer to the physical, psychological, social, or organizational aspects of the job that are able to reduce job demands, stimulate personal growth, learning and development and/or are functional in achieving work goals (Bakker and Demerouti, 2007). Resources at the workplace play a major role in the relationship between stress and strain (Demerouti et al., 2002) and have been studied in a variety of different frameworks: the demand-control model (Karasek, 1979), the effort-reward imbalance model (Siegrist, 1996), the job demands-resources (JD-R) model (Bakker and Demerouti, 2007), and the resources/recovery-stress model (Kallus and Kellmann, 2000; Kallus, 2016). Following the demand-control model, aspects of job control (e.g., latitude, autonomy) buffer the effect of stress on strain. In the effort-reward imbalance model, the important job resource is reward (e.g., salary, promotion, job security or esteem reward) that may buffer the critical relationship between stress and strain.

The important role of resources is also illustrated in Hobfoll’s (1989) conservation of resources (COR) model. This model proposes that strain is a result of a threat to resources, actual loss of resources or insufficient gain of additional resources. Any of these three paths can cause strain and might lead to burnout over time.

In the JD-R model (Bakker and Demerouti, 2007), negative outcomes of stress can be decreased if job resources are high. In contrast to Karasek’s (1979) or Siegrist’s (1996) model, the JD-R model does not limit resources to one aspect (e.g., job control or reward) but expands the view of resources to a variety of states. Indeed, examples of job resources can be found on many levels, such as on the organizational level (e.g., career opportunities, salary), on the interpersonal level (e.g., social support from colleagues or supervisor) or on the task/work level (e.g., skill variety, participation possibilities) (Bakker et al., 2007). When a sufficient amount of job resources is available, it is even possible to have positive effects of stress in the sense of challenging jobs (Bakker et al., 2010). This again is in line with the model in the ISO 10075-1.

Kallus’ (2016) resources/recovery-stress model assumes that workplace stress can be especially harmful and lead to negative outcomes if the relation between stress and recovery/resources is imbalanced. In his view the terms recovery and resources are used as nearly interchangeably. If a person experiences high stress, this stress can lead to strain depending on the person’s individual resources and depending on the person’s recovery processes to strengthen resources. In this model, resources are able to buffer the relation between stress and strain only if they have been recovered. This view is also shared by Oerlemans and Bakker (2014), where resources have to be replenished regularly to show stress-reducing effects. Therefore, the availability and state of resources must increase to the same extent as stress to cope successfully with the situation. This is in line with the COR model, where negative outcomes occur if work demands are high and resources have not been adequately replenished (Freedy and Hobfoll, 1994). Especially resources on the task level (latitude and autonomy) and on the interpersonal level (social support) have been found to reduce the negative effects of stress on strain (Halbesleben, 2006; Nahrgang et al., 2011). In the present study, we conceptualize workplace resources as a combination of task level and interpersonal level resources.

The research reported in this article used a longitudinal research design to investigate the relationship between stress, workplace resources and outcomes of stress. To measure stress, we use the concept of the areas of worklife (Maslach and Leiter, 2008), which is a framework that compasses six different stressors (workload, control, reward, community, fairness, and values). Stress is often measured as a total value by summarizing many workplace stressors (e.g., as “demands,” Hakanen et al., 2008; Trépanier et al., 2014). In our study we want to measure stress deeper in different facets; therefore, we focus on stressors described in the concept of the areas of worklife. In addition, we analyze the effects of stress at two time points (after 6 months), thus testing the causal relationship between stress and strain.

According to the ISO 10075-1 (ISO, 2000a), the immediate outcome of stress would be strain. However, we are more interested in the long-term effects of stress; therefore, the concept of burnout is used to operationalize the outcome of stress, which is a result from repeated exposure to strain (ISO, 2000a; Demerouti et al., 2002).

On basis of the studies conducted by Leiter and Maslach (2005, 2009) regarding the areas of worklife, following hypotheses are stated:

In line with the JD-R model and the resources/recovery-stress model, we hypothesize that workplace resources are negatively related to burnout and are able to buffer the negative effect of stressors (measured with the six areas of worklife) on burnout.

The main strengths of the present study lie in the differentiated view of stress by investigating the effects of different stressors and in the longitudinal design to detect causal relationships. In addition, moderating effects of workplace resources are tested cross-sectionally and longitudinally to gain better insight in the short-term and long-term buffering role of workplace resources in the stress–strain relationship.

The data were collected as part of a larger longitudinal study conducted among Austrian workers. For this study, persons were recruited from other studies executed at the Department of Psychology at the University of Graz and were asked for their consent to contact them again for future studies. The approval of the ethical commission was obtained before the start of the whole longitudinal study. The first measurement took place in spring and the second in autumn. The measurement times were carefully chosen to avoid holiday seasons. The time interval between the single measurement points was 6 months. The two measurement points are referred to as Time 1 (T1) and Time 2 (T2).

At T1, 626 participants filled-in all questionnaires in the online survey. At T2, 439 participants took part. The final sample consisted of 141 participants that took part at both measurement points and also were at the same workplace at both times. Of these 141 respondents, 61% were female (male: 39%) and their mean age was 43.7 years (SD = 9.04). The majority worked full-time or more (83.7%), 16.3% worked part-time. The participants worked in different industrial sectors, mostly from health sector (19.9%), manufacturing (12.1%), public sector (11.3%), and general services (9.2%). To analyze a possible drop-out bias, participants who participated in T1 were compared to those who participated in both waves. The analysis revealed that both groups did differ significantly in their experience of workplace resources (p < 0.05). More specifically, workplace resources were higher in the group that participated in both waves compared to participants who participated only in T1. For all other variables, no significant effects were found.

With regard to analyses, we used bivariate correlation and hierarchical regression analyses. Hierarchical multiple linear regression analyses were performed for each dependent variable (emotional exhaustion, cynicism, personal accomplishment) and were conducted for the cross-sectional sample (at T1) and the longitudinal sample. In the longitudinal analysis, the dependent variable at T1 was controlled for all outcomes. To analyze the interaction between the variables, the interaction terms first were mean-centered (z-transformed) to minimize multicollinearity and then entered in the regression analysis. The data was analyzed using SPSS Version 22.

Descriptive statistics (means and standard deviations) all study variables are shown in Table 1. The descriptive statistics were calculated for both measurement points separately. The data for workplace resources and the three burnout dimensions were compared to the data of a representative Austrian sample, collected in 2015 (Jiménez et al., 2016b). Compared to the representative Austrian sample, the study sample showed higher values in workload, emotional exhaustion and cynicism and lower values in fairness and workplace resources (see Table 1). Intercorrelations between variables for T1 and T2 (cross-sectional and longitudinal) as well as internal consistencies (Cronbach’s Alpha, α) are shown in Table 2.

To test the proposed hypotheses, hierarchical multiple linear regression analyses were conducted for the cross-sectional sample (at T1) and the longitudinal sample. In Table 3, the results for T1 are depicted. In the first step, the areas of worklife were stepped into the equation, which was significant for all models (ΔR²_{emotional exhaustion} = 0.55; ΔR²_cynicism = 0.57, ΔR²_{personal accomplishment} = 0.37). The second step included workplace resources, which significantly accounted for an additional variance for all outcomes (ΔR²_{emotional exhaustion} = 0.04; ΔR²_cynicism = 0.05, ΔR²_{personal accomplishment} = 0.07). The third step included the interaction terms, which were non-significant for all models.

Out of the areas of worklife, the areas of reward and values showed to be the most important predictors for all three dimensions of burnout. The area of workload was significant for the criteria emotional exhaustion and cynicism. Workplace resources were negatively related to emotional exhaustion and cynicism and positively related to personal accomplishment. As for the interactions between the areas of worklife and workplace resources, only the interaction between workload and workplace resources was significant for the outcome variable emotional exhaustion. More specifically, the buffering effect of resources seems to work better in an environment with low workload, as low workload combined with high resources has the lowest level of emotional exhaustion (Figure 1).

The standardized regression coefficients (β) as well as p-values and ΔR² are presented in Table 3.

Table 4 present the four steps of the analyses. The dependent variables at T1 were stepped into the equation first. This step was significant for all models (ΔR²_{emotional exhaustion} = 0.54; ΔR²_cynicism = 0.43, ΔR²_{personal accomplishment} = 0.44). This step was followed by the six areas of worklife (step 2). The first step accounted for an additional variance for cynicism (ΔR²_cynicism = 0.06). The third step included workplace resources, which was non-significant for all outcomes. In the fourth and last step, the interaction terms were included. This step was not significant for all outcomes.

Workload showed to be the most important predictor for emotional exhaustion at T2 and reward was an important predictor for cynicism at T2 (hypotheses 1 and 2). The other areas of worklife did not show significant results with all three dimensions of burnout at T2. Contrary to hypothesis 3, resources at the workplace did not show any direct relationships with all three outcome variables.

Out of all interactions, only one was significant: the interaction between workload and workplace resources was significant for personal accomplishment. More specifically, high workload paired with high workplace resources at T1 lead to higher personal accomplishment at T2 (Figure 2). As for emotional exhaustion and cynicism, none of the proposed interactions showed significant results.

The standardized regression coefficients (β) as well as p-values and ΔR² are presented in Table 4.

An important point of this study lies in the investigation of different types of stress (stressors) instead of using one global indicator of stress. By analyzing different stressors together with different types of resources, the complex interaction between stressors and resources could be investigated much more deeply. In the present study, we used a global indicator of resources as the questionnaire used in this study (RESTQ-Work-27; Jiménez et al., 2016a) only allows building a global score of resources. In future studies, different types of stressors and resources should be assessed to get a deeper understanding in the complex interactions at the workplace.

Same-source bias is a possible limitation of the study. We collected data at different points in time, a practice suggested by Podsakoff and Organ (1986) to reduce the effects of common method bias. Nevertheless, it is possible to take objective indicators for work environment for the analyses (e.g., sickness absences, accident statistics), but these are usually difficult to obtain as companies are hesitant to hand over sensitive data.

As the online survey was declared as “stress study” and open for every interested person, a self-selection bias could occur. Comparing the values of the study sample with a representative sample of Austrian workers, the study sample consists of persons with more burnout and higher stress. This “ceiling effect” has not just drawbacks but also “advantages” in the more conservative sense of research: It is even more difficult to detect significant effects if a ceiling effect is present.

A reason for not finding more longitudinal effects may lie in the time interval of 6 months. It seems plausible to assume that for a health indicator such as burnout, which is considered as relatively stable, permanent changes are not easily achieved but require an exposure over a longer time interval. This assumption is supported by other analyses where effects over a period of 1 year were found (Maslach and Leiter, 2008). Non-significant results also could be due to lack of power because of our small sample size. We therefore emphasize that some coefficients are not significant yet substantial in absolute value, so we do not claim that there is no effect in this case.

The findings in the presented study show small effects for the buffering effect of workplace resources on the relationship between stress and the long-term consequence, burnout. In the cross-sectional sample, workplace resources seem to be important prevention factors for burnout, but moderating effects with stressors are still weak. As for the longitudinal effects, more data are needed to investigate this relationship further. Our study includes effects after half a year, but extending the research for longer time periods is needed.

Nevertheless, we stress that workplace resources are important and so it has to be looked on short- and long term effects of resources. In the practical field of risk assessment, assessing workplace resources together with stressors is sometimes overlooked but important to develop sustainable interventions. The stress-strain chain which is the base of risk analysis and the design of workplaces (ISO, 2000b) has to include the stressors as well as buffering aspects especially (Portuné, 2012; Zoni and Lucchini, 2012). For the risk of burnout – which is seen as a long-term impairing effect of stressors (Demerouti et al., 2002; ISO, 2015) – especially the time aspect must be considered in combination with different stressors. Looking at workplaces with high workload for sure the rule of work design means that at first the work environment has to be changed (Portuné, 2012). Especially for these workplaces possible workplace resources should be taken into account when workplace design is considered.