This thesis explored the ability of two trait EI subscales [Emotional recognition and expression (ERE), and Emotional control (EC)] to explain significant amounts of unique variance in health variables. It asked first, whether the relationship between trait EI and health was mediated by coping, social support or unhealthy behaviours; and second, whether the harmful effect of stressor exposure on health was moderated by trait EI subscales. The thesis focussed on two specific components of EI to aid understanding of how specific elements of trait EI influence health, cross sectional and longitudinal designs were used; both objective (salivary cortisol) and subjective (life event inventory) measures of stress were used; personality, gender and age were considered as control variables wherever the predictive power of EI was explored, and health was explored as a multidimensional construct. Additionally, the selected trait EI measure [the Swinburne University Emotional Intelligence Test (SUEIT; Palmer & Stough, 2001)],was well matched to the ability EI model proposed by Mayer & Salovey (1997) and did not include correlates of trait EI such as facets personality. Original contribution to knowledge are; first, the longitudinal investigation of trait EI subscales and health; and second, the exploration in a naturalistic setting of the capacity of trait EI subscales to explain significant variance in cortisol reactivity, when personality, gender and age were controlled.
Results revealed neither ERE nor EC could explain significant amounts of variance in health variables (cross-sectionally or longitudinally), or in cortisol reactivity. However both ERE and EC were found to moderate the relationship between life event stressor exposure and health status. Moderational analyses revealed that, under a high frequency of stressful events, health was worse when EI subscales were low. In combination the results of these studies suggest that trait EI subscales ERE and EC are predictive of health only under high stress conditions. This finding is contradictory to the findings of recent meta analysis (Martins, Ramalho & Morin, in press), and discussion suggests that the discrepancy may be because past studies have used trait EI measures with content wider than the ability EI model (such as personality and happiness), which increased predictive power but reduced theoretical understanding.