Managing work-life policies: disruption versus dependency arguments. Explaining managerial attitudes towards employee utilization of work-life policies

Abstract

What factors shape managerial attitudes towards the utilization of work-life policies? The influence of disruptiveness (Powell and Mainiero 1999) and dependency (Klein, Berman and Dickson 2000) arguments on managerial attitudes is examined using a vignette design. In this study, managers in four financial firms in the UK and the Netherlands were asked to judge hypothetical requests from employees to utilize work-life policies. Findings indicate that managers are mainly interested in the performance of their department or work unit, and work-life policies are often seen as disruptive. However, when the least disruptive request is considered (short-term leave), we find that dependency arguments also play an important role and managers are more likely to respond to their employees' personal and family needs.

Keywords:

• dependency
• managerial attitudes
• vignette study
• work-life balance
• work disruption
• work-life policies

Notes

1. In the UK, employees are entitled to 26 to 52 weeks (depending on tenure) maternity leave, 2 weeks paternity leave and 13 weeks parental leave. In the Netherlands employees are entitled to 16 weeks maternity leave, 2 days paternity leave and 13 weeks parental leave (e.g. den Dulk and van Doorne-Huiskes 2007).

2. A “nested” data structure means that one unit at the lower level (i) falls under one unit at the higher level (j). For the present study, this means that each request is assessed by only one manager (no two managers ever assessed precisely the same request), given that each manager was asked to bear in mind the characteristics of his/her own department as well as its most common position when considering the requests. Since each manager works in a different department, all requests are consequently “unique”.

3. In a “regular” regression model, the fact that two (independent) populations – one of managers and one of requests – were involved would be ignored (Woodhouse, Rasbash, Goldstein and Yang 1995). Regression models ignore that the data consist of several levels and that n is overestimated. This makes the conclusion more likely that manager characteristics have significant effects on the assessments of different requests, whereas in reality there are no such effects (Type I error).