Doctor of Philosophy / Grossarth-Maticek and Eysenck (1988) demonstrated that personality type as measured by the 70-item Grossarth-Maticek Personality Stress Inventory (GMPSI), predicts mortality and its cause with remarkable accuracy. Further, various forms of autonomy training, such as bibliotherapy and short individual treatment that were designed to reduce the toxic personality features (such as emotional dependence) effectively improved long-term health outcomes. However, several aspects of their theory and research were either insufficiently explicated or require further investigation. For example, the researchers did not thoroughly investigate the psychometric properties of the inventory, and they did not sufficiently examine the mechanisms that may mediate the personality-health relationship. Further, the autonomy training was inadequately described, and perhaps could have been briefer and had the same impact. Three studies were run to investigate these issues further. Study 1 (Chapter 2) investigated the internal consistency, test-retest reliability and convergent, discriminant and concurrent validity of the GMPSI in 312 first-year students, using a variety of statistical techniques (e.g., structural equation modeling, confirmatory factor analysis, and simple correlations). In addition, the psychometric properties of this scale were compared to those of three revised versions of the scale, which were theorised to be psychometric improvements of the original version from basic principles. The revised versions contained reverse-worded items (of different types) to monitor and disrupt acquiescence response sets, and/or an extended response scale to improve internal consistency and stability. Psychometric effects of reverse-wording and the extended response scale on scale reliability were examined via the analysis of construct reliability estimates, personality subscale model fit of congeneric measurement models (a form of structural equation modeling), and test-retest reliability estimates. The original version of the scale appeared to be the most reliable and valid scale of the four versions. This improved reliability of the original version was not an artifact of an acquiescence response set, because this form of responding was not prevalent in the reverse-worded versions. In contrast, it appeared that the incorporation of reverse-wording degraded the internal consistency of the scale as participants appeared to respond to the positive and negative-worded items as if they were measurements of independent constructs, rather than measurements of constructs on opposite ends of a particular dimension. Predicted correlations between the GMPSI and concurrent validity measures offered support for Grossarth-Maticek's theory, and suggest that the GMPSI is an effective and reliable tool for the measurement of these personality types. However, experimental evidence supporting the link between personality, mediating mechanisms and disease is required to further substantiate these findings. The second and third studies (Chapters 3 and 4, respectively) were investigations of the utility of an information pamphlet discussing stress, assertiveness and relaxation, in improving GMPSI personality type stress responses in a student and a community sample (Studies 2 and 3 respectively), and provided an opportunity to investigate the mechanisms mediating the personality-health relationship via experimental manipulation. Specifically, Study 2 investigated the effectiveness of the pamphlet in improving personality scores, mood, coping strategies, health behaviours, and salivary cortisol levels (measured at pre-treatment, 1-month post-treatment and 5-months post-treatment) in 200 first-year university students. Further, two modes of administration of the pamphlet were compared: pure selfadministration versus instruction accompanied by self-administration. Group differences in mood and cortisol reactivity to a visualisation stress task were also assessed. Very few significant differences were found between the two pamphlet administration forms. Both pamphlet forms were effective in improving mood states, some lifestyle habits (e.g., exercise), and salivary cortisol responses to the acute stress task compared to the control group. The strength of these effects ranged from small to medium, and all significant differences were between pre-treatment and the 1-month post-treatment session. The failure to observe differences at the 5-months post-treatment session may have been due to inadequacies of the treatment, or low statistical power for detecting effects from the final session due to the large attrition rate that had occurred by this session. Overall, while large treatment effect sizes were not found, the results could be construed as "clinically" significant when taking into account the low costs of implementing an information pamphlet in the larger community, and the potential benefits on individuals' stress responses and health behaviours. Study 3 aimed to investigate the treatment effects of the same self-administered pamphlet in 77 participants from the general community, who varied largely in age (19 - 77 years). Treatment and control groups were compared in terms of treatment compliance, cortisol levels, health behaviours, personality scores, perceived stress, mood, and coping styles (measured at pre-treatment, 2 weeks posttreatment, 3 months post-treatment, and 6 months post-treatment). In addition, group differences in cortisol reactivity to an acute cognitive stress task were examined. Home visits were arranged for each session to reduce sample attrition. Further, an intervention evaluation form was administered at each post-treatment session to verify and maintain treatment involvement. Several significant treatment effects were observed, including changes in personality scores and non-productive coping strategies, and the strength of these effects ranged from medium to very large. Most participants reported that they found the information pamphlet very appealing and helpful. In addition to the examination of intervention effects in the second and third studies, a path model that aimed to identify direct and mediating relationships between personality and concurrently measured disease was examined for the two samples (Chapter 5). This path model was based on a new integrative theory of personality-disease, which was developed. Eysenck's (1991) proposed personality-disease model formed the foundation of this new theory, and elements of several other generic personality-disease models were also incorporated. This theory was empirically tested using path analysis on the student and community data separately. Common pathways in the two models were then tested for invariance. Overall, most paths proposed by the integrative model were identified in one or both samples; thus, the model was generally supported. All common direct paths were statistically invariant (i.e., equivalent) in the two samples. While personality did not appear to directly predict illness, several significant indirect pathways were identified by which personality appears to affect disease incidence, such as via perceived stress, mood, coping styles, and physical risk factors. These findings appear to support Grossarth-Maticek's theory that personality affects disease incidence via stress responses, as well as other generic approaches (i.e., the personality-induced hyperreactivity model and the stress moderator model) that emphasise the stress-moderating effects of personality on health. However, support for the dangerous behaviours model was not found, which posits that certain personality dispositions seek risky behaviours (e.g., poor health behaviours such as smoking and alcohol consumption) that fit their personality. The final Chapter 6 directly compares the results of these three studies, and discusses their practical and theoretical significance in terms of Grossarth-Maticek's theory and research, views of critics, and other personality-health perspectives and research. In summary, the current studies appear to suggest that the GMPSI is a reliable and valid scale for the measurement of particular personality traits. Further, there appears to be evidence to suggest that personality traits can be changed by an intervention pamphlet. This information pamphlet also appears to be effective in significantly improving responses to stress, and these effects are more prominent in high-risk groups (i.e., subjects with extreme personality trait scores). Moreover, there is some evidence to suggest that personality may have direct effects on several mechanisms involved in the development of disease. Overall, this thesis demonstrates the importance of recognising the role of personality and stress in disease prevention and prediction by providing independent evidence for the benefits of treatment and mechanisms by which benefits may occur.

Doctor of Philosophy / Grossarth-Maticek and Eysenck (1988) demonstrated that personality type as measured by the 70-item Grossarth-Maticek Personality Stress Inventory (GMPSI), predicts mortality and its cause with remarkable accuracy. Further, various forms of autonomy training, such as bibliotherapy and short individual treatment that were designed to reduce the toxic personality features (such as emotional dependence) effectively improved long-term health outcomes. However, several aspects of their theory and research were either insufficiently explicated or require further investigation. For example, the researchers did not thoroughly investigate the psychometric properties of the inventory, and they did not sufficiently examine the mechanisms that may mediate the personality-health relationship. Further, the autonomy training was inadequately described, and perhaps could have been briefer and had the same impact. Three studies were run to investigate these issues further. Study 1 (Chapter 2) investigated the internal consistency, test-retest reliability and convergent, discriminant and concurrent validity of the GMPSI in 312 first-year students, using a variety of statistical techniques (e.g., structural equation modeling, confirmatory factor analysis, and simple correlations). In addition, the psychometric properties of this scale were compared to those of three revised versions of the scale, which were theorised to be psychometric improvements of the original version from basic principles. The revised versions contained reverse-worded items (of different types) to monitor and disrupt acquiescence response sets, and/or an extended response scale to improve internal consistency and stability. Psychometric effects of reverse-wording and the extended response scale on scale reliability were examined via the analysis of construct reliability estimates, personality subscale model fit of congeneric measurement models (a form of structural equation modeling), and test-retest reliability estimates. The original version of the scale appeared to be the most reliable and valid scale of the four versions. This improved reliability of the original version was not an artifact of an acquiescence response set, because this form of responding was not prevalent in the reverse-worded versions. In contrast, it appeared that the incorporation of reverse-wording degraded the internal consistency of the scale as participants appeared to respond to the positive and negative-worded items as if they were measurements of independent constructs, rather than measurements of constructs on opposite ends of a particular dimension. Predicted correlations between the GMPSI and concurrent validity measures offered support for Grossarth-Maticek's theory, and suggest that the GMPSI is an effective and reliable tool for the measurement of these personality types. However, experimental evidence supporting the link between personality, mediating mechanisms and disease is required to further substantiate these findings. The second and third studies (Chapters 3 and 4, respectively) were investigations of the utility of an information pamphlet discussing stress, assertiveness and relaxation, in improving GMPSI personality type stress responses in a student and a community sample (Studies 2 and 3 respectively), and provided an opportunity to investigate the mechanisms mediating the personality-health relationship via experimental manipulation. Specifically, Study 2 investigated the effectiveness of the pamphlet in improving personality scores, mood, coping strategies, health behaviours, and salivary cortisol levels (measured at pre-treatment, 1-month post-treatment and 5-months post-treatment) in 200 first-year university students. Further, two modes of administration of the pamphlet were compared: pure selfadministration versus instruction accompanied by self-administration. Group differences in mood and cortisol reactivity to a visualisation stress task were also assessed. Very few significant differences were found between the two pamphlet administration forms. Both pamphlet forms were effective in improving mood states, some lifestyle habits (e.g., exercise), and salivary cortisol responses to the acute stress task compared to the control group. The strength of these effects ranged from small to medium, and all significant differences were between pre-treatment and the 1-month post-treatment session. The failure to observe differences at the 5-months post-treatment session may have been due to inadequacies of the treatment, or low statistical power for detecting effects from the final session due to the large attrition rate that had occurred by this session. Overall, while large treatment effect sizes were not found, the results could be construed as "clinically" significant when taking into account the low costs of implementing an information pamphlet in the larger community, and the potential benefits on individuals' stress responses and health behaviours. Study 3 aimed to investigate the treatment effects of the same self-administered pamphlet in 77 participants from the general community, who varied largely in age (19 - 77 years). Treatment and control groups were compared in terms of treatment compliance, cortisol levels, health behaviours, personality scores, perceived stress, mood, and coping styles (measured at pre-treatment, 2 weeks posttreatment, 3 months post-treatment, and 6 months post-treatment). In addition, group differences in cortisol reactivity to an acute cognitive stress task were examined. Home visits were arranged for each session to reduce sample attrition. Further, an intervention evaluation form was administered at each post-treatment session to verify and maintain treatment involvement. Several significant treatment effects were observed, including changes in personality scores and non-productive coping strategies, and the strength of these effects ranged from medium to very large. Most participants reported that they found the information pamphlet very appealing and helpful. In addition to the examination of intervention effects in the second and third studies, a path model that aimed to identify direct and mediating relationships between personality and concurrently measured disease was examined for the two samples (Chapter 5). This path model was based on a new integrative theory of personality-disease, which was developed. Eysenck's (1991) proposed personality-disease model formed the foundation of this new theory, and elements of several other generic personality-disease models were also incorporated. This theory was empirically tested using path analysis on the student and community data separately. Common pathways in the two models were then tested for invariance. Overall, most paths proposed by the integrative model were identified in one or both samples; thus, the model was generally supported. All common direct paths were statistically invariant (i.e., equivalent) in the two samples. While personality did not appear to directly predict illness, several significant indirect pathways were identified by which personality appears to affect disease incidence, such as via perceived stress, mood, coping styles, and physical risk factors. These findings appear to support Grossarth-Maticek's theory that personality affects disease incidence via stress responses, as well as other generic approaches (i.e., the personality-induced hyperreactivity model and the stress moderator model) that emphasise the stress-moderating effects of personality on health. However, support for the dangerous behaviours model was not found, which posits that certain personality dispositions seek risky behaviours (e.g., poor health behaviours such as smoking and alcohol consumption) that fit their personality. The final Chapter 6 directly compares the results of these three studies, and discusses their practical and theoretical significance in terms of Grossarth-Maticek's theory and research, views of critics, and other personality-health perspectives and research. In summary, the current studies appear to suggest that the GMPSI is a reliable and valid scale for the measurement of particular personality traits. Further, there appears to be evidence to suggest that personality traits can be changed by an intervention pamphlet. This information pamphlet also appears to be effective in significantly improving responses to stress, and these effects are more prominent in high-risk groups (i.e., subjects with extreme personality trait scores). Moreover, there is some evidence to suggest that personality may have direct effects on several mechanisms involved in the development of disease. Overall, this thesis demonstrates the importance of recognising the role of personality and stress in disease prevention and prediction by providing independent evidence for the benefits of treatment and mechanisms by which benefits may occur.