This seminar is hosted by the Work, Employment, People and Organisations (WEPO) Division at Sheffield University Management School.
On the relationship between quantitative and qualitative job insecurity, and well-being: curvilinear relationships?
In his renowned Vitamin Model, Peter Warr developed the hypothesis that the association of job characteristics with aspects of well-being is nonlinear. Some job characteristics are associated in a CE manner (constant effect), and others show an AD pattern (additional decrement) with well-being. Job (in)security, as part of the umbrella concept ‘environmental clarity’, is hypothesized to show an AD pattern: low as well as high levels of security are supposed to be associated with low levels of well being, whereas moderate levels are supposed to be associated with higher levels of well-being. In short: too much insecurity is ‘bad’, but so is too much security. Only ‘moderate’ levels of security are ‘good’ for health and well-being.
In this lecture, I would like to put these assumptions to a test. In doing so, I use data from a large scale online survey (N = 2 444 respondents) gathered in 2012. The sample is rather homogeneous, even though highly skilled employees are overrepresented and blue-collar workers underrepresented. Job insecurity is measured in two ways: as quantitative job insecurity (“the probability that one might lose the actual job in the near future and become unemployed, and the worries related to that threat”), and qualitative job insecurity (“the probability that valued job features might change in a negative way in the future, and the worries related the that threat”). A large variety of health and well-being variables is used as dependent variable: job and life satisfaction, work engagement, recovery need (a variable closely related to the exhaustion dimension of burnout), mental health complaints and physical health complaints. The nonlinearity of the relationships is tested by analyzing the (standardized) squared job insecurity variable in an hierarchical regression analysis, after controlling for demographic variables and its ‘normal’ (lineair) job insecurity variant.
Most results do not corroborate the hypotheses of AD shaped associations with job insecurity, even though some curvilinear associations were found. First of all, 5 out of 12 tests show a linear relationship, and 3 a CE shaped association. In only 4 out of 12 tests, an AD pattern is found. However, none of these associations showed the expected AD pattern, in which low and high values of insecurity are supposedly associated with low scores for health or well-being. Additionally, the nonlinear associations - when significant at all - explain very little of the variance in the health and well-being variables.