There have been many attempts to alleviate the high turnover of nurses, yet the problem persists. Most strategies have focussed on improving the quality of the workplace, by increasing salary, professional status, and altering organizational characteristics. However, in spite of these efforts high amounts of nursing turnover persists.

The purpose of this study was to gain further understanding of those variables that would increase our ability to predict job turnover in nursing. This study evaluated the adequacy of a model for predicting the causes of nurse turnover. The prominent feature of this model, person-environment fit, demonstrates the relationship between an individuals perceived "fit" with the working environment and their intentions of leaving or staying in that environment. The model was also intended to demonstrate the relationships between person-environment fit with job satisfaction, and job stress as an indirect effect on job turnover.

Data for this study was collected using an anonymous cross-sectional survey approach. Latent variable model analysis, as well as other multivariate techniques were used to test the hypothesized relationships of individual and organizational variables with anticipated turnover. Regression analysis showed that enjoyment of the job, staffing levels, and professional status were predictive of nurses turnover.

The results of the present study were unable to support the hypothesis model. Although some evidence exists that indicates person-environment fit has an effect on job satisfaction, the overall model did not fit the data. A trimmed model, that was found to fit the data, did not include any significance paths between person-environment fit and job turnover. Similar to previous research strong evidence was found that demonstrated the predictability of turnover by job satisfaction variables.

Results did not support contentions of previous studies by Atwood & Hinshaw (1984), that found effects on turnover by job stress, to be mediated by job satisfaction. This was primarily due to the high correlation among the job satisfaction and job stress constructs with each other. This correlation was discovered by testing a latent variable model using the EQS program.