A stressful working environment has been acknowl- edged as an emergent health issue (European Agency for Safety and Health at Work, 2010). Much evidence has been accumulated on the link between adverse psychosocial working condi- tions and a number of health and business out- comes (Bond et al., 2006), calling for stress management interventions (SMIs). To be effect- ive, SMI should target both individual and or- ganizational levels, considering the needs and capacities of both employees and the organiza- tion (Nielsen et al., 2010). Expanding SMI to the organizational-level represents an advance from
single-measure interventions to a dynamic pro- cess of organizational change, ultimately enab- ling companies to manage health themselves. Further, the exclusive focus on stressors has been expanded, including job resources as a factor that mitigates the pathogenic effects of stressors while unfolding a distinct motivational potential (Bauer and Jenny, 2012). Simultaneously, such SMI are not limited to employees with a high risk of disease and long-term sickness for whom intensive and behavior-tailored programs are derived. They target the overall working popula- tion with an average health distribution, aiming to preserve and enhance their health status. In this context, SMI follow a salutogenic approach aiming at strengthening resources and slowly but steadily building a sustainable health-promoting working environment. The literature on the effectiveness of SMI shows that employees benefit from individual-level SMI (Richardson and Rothstein, 2008). However, there is still a lack of studies evaluating the effects of organizational-level interventions and the results of the few existing ones vary (LaMontagne et al., 2007). A systematic review by Bambra et al. (Bambra et al., 2007) found that some of the par- ticipatory organizational-level interventions improved employee health problems such as general complaints, emotional exhaustion and musculoskeletal disorders by increasing job control. Bond et al. (Bond et al., 2006) also found significant effects of organizational-level inter- ventions on business outcomes (i.e. decreased absenteeism, lower staff turnover, better object- ive and subjective performance ratings). In sum, organizational-level interventions have the po- tential to produce positive effects, but they appear to show diverse and partially contradict- ing results in terms of the combination of inter- vention elements and effect magnitudes over a range of outcomes. This is attributed to the het- erogeneity of the studies in terms of sample sizes, time lags, intervention components, effect mea- sures, study context, etc. [cf. (Biron et al., 2012)]. Further, the dynamics of change in organizations limit the ecological validity of predefined inter- ventions implemented under controlled condi- tions in pre-assigned intervention and control groups. Moreover, such research is often limited to short follow-up periods such as 3–12 months (Zapf et al., 1996), leaving long-term effects as well as the routinization of the interventions unnoticed [cf. (Taris and Kompier, 2003)]. As SMI are interventions into complex social systems, Kompier and Kristensen (Kompier and Kristensen, 2000) acknowledge that most SMI studies require non-traditional research designs. Semmer (Semmer, 2006) argues in favor of chan- ging the focus from outcome variables to work characteristics as determinants of health and well- being, and notes that more detailed analyses and documentation of context and process factors in- fluencing intervention success are needed, rather than simply criticizing (supposedly) poor designs. This view is also echoed and advanced by Nielsen and Randall (Nielsen and Randall, 2012), who include information on the intervention process as a relevant moderating variable explaining the variance in the effectiveness of the intervention.
Randall et al. (Randall et al., 2005) argued that measurement of the intervention process can be used to adapt and shape the design of the effect- iveness evaluation: Intervention exposure and ap- praisal serve to retrospectively assign employees to intervention and control groups and thus support quantitative outcome evaluations where controlled quasi-experimentation is not possible (Randall et al., 2005). If such ‘less-than-optimal’ designs are applied, then strong theoretical inter- vention evaluation frameworks (Chen, 1990) and mixed methods (Leech and Onwuegbuzie, 2009) should be used to capture the intervention context and process in order to plausibly attribute observed effects to the implemented intervention and cross-validate the results