ABSTRACT
This day-level study examined the role of perceived organizational support (POS) in the context of employees’ negative work reflection during off-job time. We hypothesized that negative work reflection during off-job time should be indirectly related to reduced work engagement on the next workday through personal resources (i.e., vigour and self-efficacy) in the morning. In addition, we hypothesized that POS moderated the relationships between negative work reflection and personal resources and between personal resources and work engagement. In total, 100 employees completed one general survey and three daily surveys (in the morning, after work, and at bedtime) over five workdays. Results of multilevel path analyses showed that negative work reflection was neither directly associated with personal resources nor indirectly with work engagement via personal resources, although vigour and self-efficacy positively predicted increased work engagement. However, negative work reflection was negatively associated with self-efficacy when POS was low. POS did not predict work engagement, but moderated the relationships between personal resources and work engagement: Consistent with the resource substitution hypothesis, high levels of POS compensated for low levels of vigour and self-efficacy. Negative work reflection had a significant negative indirect effect on work engagement through self-efficacy only when POS was low.
Acknowledgment
This study is part of Anna R. Ott’s (née Koch) dissertation. We thank Carolin Boye and Vanessa Preuss for their support during data collection.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. This is the first publication from this data set.
2. In addition, in our study, employees were nested in organizations. However, as ICCs showed that less than 4% of the variance in all day-level variables was attributable to the organization level, we specified a two-level model (with days nested in persons) only.
3. As the models initially specified with random intercepts and slopes did not converge, we set the residual variances of the random slopes to 0. This procedure implies that all the slope variance is accounted for by the Level 2-predictor POS.