|The program of research presented in this thesis focuses on the construct of employee turnover. It is argued, despite the intense interest that employee turnover has attracted (Cotton & Tuttle, 1986; Griffeth, Hom & Gaertner, 2000), that turnover theory and research suffers from a number of notable shortcomings. These shortcomings primarily stem from the development and study of employee turnover models that attempt to make predictions for all individuals, at all times and across all places (Lee & Mowday, 1987). It is proposed that three shortcomings need to be addressed, specifically the need to develop employee turnover models that are organisationally specific, including predictors that are particularly relevant to the context under study and consideraton of the impact of career stage and gender on this process. Given the argument that employee turnover models need to be developed according to organisational context, a specific organisational population was identified for study in this research program. The current research program focused on a sample population of police officers employed in an Australian police organisation. This population was chosen due the paucity of research that has investigated turnover and turnover intention processes within the policing context and in light of the practical implications,
specifically the significant financial cost, associated with police turnover. In addition to addressing the three key shortcomings of employee turnover theory and literature as identified above, the current program of research also addressed a number of limitations that are specifically relevant to police turnover research. This research represents an important contribution to the existing body of police specific turnover research due to the limited attention that the study of employee turnover within the policing context has attracted. Furthermore, the current research program studied large sample populations of Australian police officers given that previous police turnover research has been limited by the use of relatively small sample sizes and have focused primarily on the study of officers employed in United States police organisations. This research, in contrast to existing police turnover research, utilised statistical techniques that allowed multi-variable process models of turnover intention to be developed and tested. Lastly, this research clearly articulated the type of turnover being examined. This is particularly important given that many studies of police turnover have failed to specify the specific form of turnover that is under study. This research has contributed to current understanding of voluntary resignation and voluntary turnover intention processes. Three studies comprised the empirical component of the current program of research. The series of studies that were conducted were founded on a career development sequence. Each study contributed to the overarching aims of the research program, namely to examine turnover and turnover intention processes of male and female police officers at distinct phases of the employment relationship. Study One examined turnover at the earliest stage of the employment relationship, the recruitment phase. This study tested the hypothesis that personality is predictive of voluntary resignation of male and female police officers. Using logistic regression analysis, the results of this study indicated that male and female police officers (N = 560) who voluntary resign from the police organisation compared to those who remain, are more affected by feelings than emotionally stable, are more tender-minded than tough-minded and are more venturesome than shy. No evidence was found to suggest that differential predictive relationships exist between personality and turnover as a function of gender. Study Two examined the turnover intention processes of police recruits (Time One: N = 253; Time Two: N = 253; Time Three: N = 216) engaged in their first year of employment with the police organisation. Using multiple group structural equation modelling (multiple group SEM), it was concluded that socialisation stage and gender impacts on the relevance of predictors contained within turnover intention models and gender affects the strength of predictor relationships. Moreover, this study provided evidence to suggest that turnover intention process models should include variables that are particularly relevant to the organisational context being studied. This study found that two variables identified as particularly relevant to police organisations, namely person-organisation and person-job fit, play key roles in police specific turnover intention processes. Study Three examined the turnover intention processes of male and female police officers (N = 996) across the career span. Using multiple group SEM, it was concluded that career stage impacts on the relevance of turnover predictors and in some instances, career stage and gender impacts on the strength of predictor relationships. More specifically, it was concluded that a single model largely captures the process of turnover intention for officers in early and middle career stages while a slightly altered model is needed for officers in the late career stage group. Furthermore, this study suggested that organisationally relevant predictors of turnover intention, in this instance, police specific stress, organisational stress and health status, should be included in police specific process models of turnover intention. Based on the three studies conducted under the auspices of the current program of research a number of important conclusions were drawn. It was concluded that organisational context should be considered when developing and applying models of turnover and turnover intention to Australian police organisations. In particular, an analysis of the organisational environment should be undertaken in order to identify factors that may be specifically relevant to understanding turnover and turnover intention in the context under study. Moreover, understanding of employee turnover and turnover intention is enhanced through the development of models that are tailored to specific points within the career span. Lastly, it was concluded that separate models developed according to gender are not warranted however the differential strength of predictive relationships as a function of gender should be considered.