*4.2. Limitations and Future Directions*

Several of this study's limitations provide useful directions for future research. First, we used self-reports of workplace bullying, work-to-family conflict, quality of life, and occupational health that may pose a risk for common-method bias [60]. For example, an employee who experienced more workplace bullying might have responded negatively to the items of quality of life and occupational health. Future research may benefit from incorporating objective measures of well-being, such as clinical health measures or biomarkers of stress. Second, our sample was purposely selected from multiple worksites in three industries (healthcare, education, and banking) in South Korea, and thus it is not representative of Korean employees. In the future, it is necessary to include workplace bullying items in a national survey so that we can draw national-level inference about the negative influence of workplace bullying. It may also be that our measure of workplace bullying may not fully capture the real phenomenon of workplace bullying. According to Seo [5], only about 38% of victim employees in Korea report the incidents of bullying, because of their perception that some extent of bullying is unavoidable in Korean workplace culture. As such, we may underestimate the extent of workplace bullying. Future research may need to improve the validity of workplace bullying measure by cultural and occupational contexts. More specific measurements about workplace bullying are also needed. For example, there may be differences between men and women in the experience of workplace bullying consequences of it [61]. Moreover, more regional analyses are needed to see whether findings from our study are replicated in other settings. Finally, our cross-sectional analyses cannot determine

the direction of effect. Although our analytic models imply that workplace bullying is a predictor, work-to-family conflict is a mediator, and quality of life and occupational health are outcomes, there is no temporal order between the variables and causality can operate in other directions. Future research should include multiple time points to identify the direction of effect.
