Specialeforsvar ved Anne Helby Petersen

Titel: Exploring the causal relationships between gender segregation, work environment and ill health - a statistical analysis using structural equation models

Abstract:
This project investigates the causal relationships between psychosocial work environment, occupational gender segregation and ill health. While it is generally accepted that work environment and gender segregation are both causes of ill health, no empirical studies have investigated if the effect of gender segregation is mediated through an effect on work environment. This project sets out to answer that question.

The statistical analysis is conducted within the framework of structural equation models, which allows us to formalize causal assumption into a statistical model that can be estimated using empirical data. Thus, given theoretical, causal assumptions, we have a stringent method for deducing the empirical consequences, even if the data is observational.

The analysis was based on survey data from 3233 Danish employees sampled among typical Danish occupations. Several different versions of the questionnaire were used and for investigating the implications of these variations, I have developed a new method denoted principal component analysis-based data structure comparison (PCADSC). This method allows questions about similarities in data structures to be answered through visual inspections without any parametric assumptions. Though the method needs to be inspected more closely, it shows promising preliminary results in identifying differences in data structures.

In the empirical analysis of this project, I find that the effect of gender segregation on health is almost exclusively mediated through work environment. The generalizability of this result is somewhat limited due to the nature of the data on which it is build, but nonetheless, we find that linking gender segregation with work environment proposes a promising next step in work life research.

Vejledere:  Susanne Ditlevsen / Karl Bang Christensen, SUND
Censor:      Per Bruun Brockhoff, DTU