Gwendolyn Paige Watson, MS, Clemson University Robert R. Sinclair, PhD, Clemson University

In a review of the Job Demands-Resources (JD-R) model, Bakker and Demerouti (2017) suggested future research continue to expand the model; specifically, they proposed that job crafting should be included in the model. Our study responds to this call by testing job crafting behaviors as part of a feedback loop in the job resources path of the model. Additionally, we tested competing models of the directional relationship between job crafting and work engagement.

The current study examines how job resources (e.g., work-related support resources – perceived organizational support, perceived supervisor support, and perceived coworker support) initiate motivational processes in the JD-R model that lead to employee job crafting. Job crafting occurs when employees redesign their experiences at work (Wrzesniewski & Dutton, 2001). Aligned with the JD-R model, Petrou et al. (2009) suggests job crafting may occur through seeking resources (e.g., autonomy, social support). Thus, by definition, we expect job crafting will initiate a feedback loop that increase work-related support resources. Furthermore, we confront the conflicting findings concerning work engagement and job crafting. Examination of cross-sectional, meta-analytic, and longitudinal studies emphasizes the inconsistencies in the literature about the nature of the relationship between work engagement and job crafting. Given the support for unidirectional causal pathways from both work engagement to job crafting and job crafting to engagement, we tested for reverse causation effects of work engagement and job crafting. The hypothesized relationships are shown in Figure 1.

The data consisted of two waves, six weeks apart, administered through Amazon?s Mechanical Turk (MTurk). The survey was posted to MTurk and available to all US members. Participants were rewarded $4 for the completion of each survey. A total of 439 participants completed both waves of the study and passed the required attention checks. The participants were on average 36.80 years of age (SD = 11.04), consisting of 53.2% men. Most of the participants were employed full-time (89.3%) and represented all of the occupations from the O*Net standard occupational job classification categories. The measures used in this study are provided in Table 1. We used SEM with the lavann package in R to test the hypothesized effects in addition to several alternative models per SEM best practices.

Means, standard deviations, internal consistencies, and bivariate correlations are reported in Table 1. The correlations supported the hypothesized relationships. As shown in Table 2, all of the models tested were good fit (Hu & Bentler, 1999); however, the basic JD-R model (alternative model 3) was the best fit model. The results supported Hypothesis 2 (b = .23, SE = .07, p = .002). The autoregressive paths were also significant as shown in Figure 2. Given the strong support for the JD-R model in the literature, we were surprised to not find significant results for the basic JD-R relationships (work engagement mediates resources and job crafting). Thus, we tested the expected mediation cross-sectionally for both T1 and T2 as well as a time-lagged model. All three mediation analyses demonstrated support for the basic JD-R model prediction that engagement mediated the relationship between work-related resources and job crafting. This suggests that our data are not problematic, rather the hypothesized relationships do not hold in a longitudinal design.

The findings of this study shed light on important considerations when testing job crafting in the JD-R model and contribute to understanding the nature of the relationship between job crafting and work engagement. Although we did not find evidence of the job crafting feedback loop, we found consistent support for the unidirectional relationship of work engagement predicting job crafting over time despite the highly stable nature of the study variables. The finding that work engagement predicted job crafting (and not the opposite) extends the literature by contributing to the few studies that have tested the bidirectional relationship (e.g., Harju et al., 2016; Vogt et al., 2016) using longitudinal designs. One explanation for not finding a reciprocal relationship is that workers may be less likely to job craft if they are not already engaged in their work. To initiate this relationship in practice, organizations should consider interventions that focus on work engagement (e.g., resource building, leadership training) as well as job crafting training so that employees are aware of strategies for effective job crafting.

This study responded to a call for research by Bakker and Demerouti (2017) to continue improving the JD-R model by expanding the model to include less-studied variables and using better research designs. The results support a unidirectional relationship between work engagement and job crafting. This study offered several theoretical implications and considerations that can be used to develop and improve future research studies testing job crafting in the JD-R model. Future researchers should address limitations of this study by testing these relationships over different time lags, using non-self-report measures, and collecting data from other sampling methods.

Tags: Applicable to all occupations/industries, Basic research, Empirical study, Job and Task Design, Job Attitudes; Turnover; and Retention, Organization- and Job-Level Environments and Practices, Social and Organizational Environment, Theoretical exposition or development, Workplace Stress; Outcomes; and Recovery